INTRODUCTION Within the last few decades

Within the last few decades, 3D digitization technologies have been successfully applied

within the discipline of physical anthropology, allowing researchers to non-destructively record,

visualize and replicate the delicate remains of our scarce ancestral heritage. At the most basic

level, 3D digitization allows three dimensional objects to be recorded by capturing a series of

points based on the volume or surface of an object. The first instance of 3-D digitization

technology being integrated into physical anthropology utilized volumetric computedtomography

scanning to investigate the internal structures of Egyptian mummies (HardwoodNash, 1979). The

investigators of the Tyrolean Ice Man expanded upon this use of CT scanning in mummy studies

by creating a physical replica of the cranium using stereolithography (Seidler et al., 1992). This

was the first instance of the integration of CT and rapid-prototyping technologies in an

anthropological investigation (zur Nedden et al., 1994).

In contrast to capturing volumetric data using computed tomography is surface scanning,

which is achieved through such technologies as laser and white light scanning and is regularly

used in the documentation and study of fossils as well (Aiello et al., 1998). Surface scanning is

useful to the researcher interested only in the outward morphological information of the

specimen, and it also allows the object to be viewed on the computer screen without any threat of

damage through physical handling (Tocheri, 2009). Digitizing bone can be a valuable alternative

to traditional, destructive mold-making procedures when creating copies of a specimen since the

3D model created from surface scans can be physically replicated using rapid-prototyping

technologies (D’Urso et al., 2000; Fantini et al., 2008; Pérès et al., 2004; Zhang et al., 2000).

In the subfield of physical anthropology known as forensic anthropology, the

anthropologist applies the knowledge of human osteology in a medico-legal setting. This often

results in collaborations with law enforcement and medical examiners in order to achieve a

positive identification of unknown persons. However, the implementation of 3D digitization

technologies within this subfield has been relatively slow in comparison to the rapidly expanding

applications in physical anthropology and archaeology. Much of the existing literature related to

forensic anthropology details the use of laser scanning for forensic facial reconstruction (Benazzi

et al., 2010), the virtual measurement of bone (Decker et al., 2011; Ramsthaler et al., 2010) and

trauma reconstruction (Thali, 2003). Although these applications make a useful argument, they

are not substantial enough to justify the incorporation of a 3D digitization unit in a forensic

anthropology lab. Because the technology can be expensive and requires training, a more robust

series of applications should be developed in order to persuade forensic anthropologists to more

readily implement these technologies in their investigative methodologies.

Research Questions and Objectives

The purpose of this thesis project is to investigate the use of a desk-top laser scanner in

documenting human skeletal remains that are associated with medico-legal investigations. It is

believed that the acquisition of 3D representations of human bone will provide a superior form of

documentation to traditional mediums such as photography. In addition to providing enhanced

documentation of materials, there are a multitude of further applications that can be achieved

with the use of the 3D file that results from the digitization process. The exploration of these

applications and the documentation of efficient scanning and data processing work flows will be

a focus of this research.

The research conducted in this project will be centered on the following question: What

are the applications of surface-digitization and visualization in forensic anthropology? In order to

address this question, this paper will compile the existing body of relevant literature, and use this

as a reference for adapting existing applications for use in forensic anthropology. The potential

applications surmised from the literature will then be demonstrated through the use of multiple

3D files created by using a laser-scanner and its proprietary software to scan and process surface

scan data of different bone types. The 3D models will then be processed using a variety of

affordable computer graphics programs in order to illustrate the applications for using the scan

data that can be useful to the forensic anthropologist.

The applications developed in this project will lay the groundwork for future exploration

of the possible applications for using scan data in forensic anthropology, and provide guidelines

on the many considerations involved in bone surface digitization projects. It is hoped that this

project will provide a practical framework for implementing this technology in a forensic

anthropology lab and encourage its use as a standard procedure in the data collection

methodologies of professional forensic anthropologists.

Thesis Outline

This thesis will be divided into six chapters. Chapter one will provide an introduction to

the research in this project. The second chapter will provide further background information

related to 3D digitization, and will present a brief literature review outlining other projects

relevant to this research. Chapter three will discuss the methods used in this project, highlighting

the technologies that will be implemented. The fourth chapter will discuss the results of the

scanning process, providing general recommendations for scanning bone and will also detail

specific problems and procedures related to particular bone types. Chapter five will focus on the

applications incorporating the 3D model that is produced from the scanning process. An outline

will be provided in the chapter, illustrating the steps necessary for processing the 3D model for

visual applications in forensic anthropology. The final chapter will serve as a discussion and

conclusion for the use of laser scanning in forensic anthropology based on the results of this

project.

CHAPTER TWO: BACKGROUND & LITERATURE REVIEW

This literature review will provide background information on previous research related

to the 3D digitization and visualization of bone. It will begin with background information on

3D-digitization, followed by a literature review organized by application. The most basic

application for using the scan data is its ability to be archived for future reference. From this file

simple measurements can be recorded. In addition, the object can be rendered in a virtual

environment allowing the user to rotate, zoom and manipulate lighting in order to highlight

specific areas of interest. Coupled with real-time visualization on the computer screen is the

ability to record screen shots for illustration purposes. Reconstruction of fragmented or missing

bone can be performed on the computer, avoiding the use of potentially destructive procedures

that use glue or modeling clay to rebuild incomplete bone. Several other applications, including

forensic facial reconstruction, trauma reconstruction and rapid-prototyping, are outlined within

this chapter for future development but will not be explored within this project

What is 3D-digitization?

3D-digitization is the acquisition of x, y, and z coordinates that represent an object in 3D

space (Bernardini & Rushmeier, 2002).There are two primary forms of 3D digitization: volume

digitization and surface digitization. Volume digitization is the acquisition of the entire structure

of an object, including outer surfaces and internal geometry. This is most commonly achieved

through the use of CT scanners. Surface digitization on the other hand, is the acquisition of only

the surfaces of the object being scanned (Bernardini & Rushmeier, 2002; Kappelman, 1998;

Tocheri, 2009; Zollikofer & Ponce de Leon, 2005). There are several technologies that can

achieve this, including laser scanning, photogrammetry and white-light scanning. This particular

project will implement the use of a laser scanner. Although this project will be focused on surface

digitization, a great deal of literature relevant to this project implements volume digitization. The

applications involved in volume digitization research of bone can often be easily adapted in the

processing of data that is the result of surface digitization. Because of this, some of the research

into the volume-digitization of bone will be included in this literature review. It would also be

useful for the forensic artist and anthropologist to be acquainted with volume digitization, should

they encounter a situation where case materials have been CTscanned.

Volume Digitization vs. Surface Digitization

There are several reasons why surface digitization using a laser scanner was chosen for

this research in lieu of volume digitization using a CT scanner. The primary reasons are in the

expense and expertise required in the operation of CT scanning equipment. CT scanning

equipment can cost millions of dollars and requires specialized training (Kuzminsky & Gardiner,

2012), whereas laser scanning equipment can cost only a few thousand dollars and requires only

minimal training (Komar et al., 2012). An additional reason for utilizing laser scanning in this

project has to do with the characteristics of the resulting data. Standard CT-scanning has a lower

resolution (Fantini et al., 2008) and decreased accuracy (Niven et al., 2009) when compared with

high-definition laser scanning equipment. This decreased level of accuracy in the 3D

representation of skeletal elements is not desirable in the documentation of case materials in

medico-legal investigations. The exception to this is Micro-CT scanning, which has vastly

superior resolution but is also significantly more expensive than standard CT-scanning equipment

(Slizewski et al., 2010). Volume data that results from CT-scanning also includes the complex

internal geometry of an object, and when working with complex anatomical structures it can

cause the file to become quite large. This can result in computer performance issues depending

on the processing power and amount of random-access-memory (RAM) afforded by the

computer’s hardware capabilities. The majority of 3D software programs are also not created to

handle highly dense 3D data, and thus importing CT-data into them can result in computer

performance issues. In general, surface scan data requires less advanced computer hardware and

can be less taxing on a system’s performance when imported into 3D software programs. A final

consideration is the fact that a laser scanner does not emit the harmful radiation of a CT scanner

(Park et al., 2006; Sholts et al., 2010).

Review of Applications

Once a 3D model has been obtained using the scanner and its processing software, there

are a number of ways in which it can be further processed for visualization by utilizing an

assortment of computer graphics programs. This review will document the relevant literature

based on visual applications that result from the post-processing of 3D scan data. Based on the

review of literature, the applications for laser scan data has been divided into the following

categories: documentation, measurement, visualization and illustration, virtual reconstruction,

trauma reconstruction, forensic facial reconstruction and rapid prototyping.

Documentation

The most basic application that can be completed with a 3D file once it has been

processed is to save it to the computer’s hard-drive (Komar et al., 2012; Kuzminsky & Gardiner,

2012). This is the basis of the 3D documentation of the scanned object, allowing it to be archived

and viewed at a later point in time (Park et al., 2006). All further applications using the 3D data

are contingent upon this feature. The representation of human remains in a 3D file format is

superior to traditional forms of 2D documentation such as digital photography, which are unable

to completely represent the complex morphology of skeletal elements such as the cranium

(Komar et al., 2012; Kuzminsky & Gardiner, 2012). Additionally, the 3D file allows the user to

rotate the object and view it from any angle (Niven et al., 2009; Tocheri, 2009) and save

screenshots using visualization software (Kuzminsky & Gardiner, 2012), without the potential

threat of damage to the original skeletal elements (D’Urso et al., 2000; Pérès et al., 2004;

Tocheri, 2009; Zhang et al., 2000). The ability to archive surface data insures that the original

state of the remains has been documented and can be reexamined at a later point in time, should

the original material be altered in any way (Tocheri, 2009). The digitization process can also be

used to circumvent distortions that result from photography which result in misrepresentations of

the original subject (Thali et al., 2003). The 3D data can also be digitally transferred to other

experts, allowing them to consult on cases without having to travel or transport the remains

(Davy-Jow et al., 2012). In addition, digital reference collections can be created and made

accessible on the internet, or stored and transported to the field on portable computers and digital

storage devices (Niven et al., 2009). Such collections can be useful for facilities and institutions

that face budgetary constraints or have limited storage space (Niven et al., 2009).

Measurement

One of the most useful advantages of having a 3D model of human bone is the ability to

take virtual measurements. Traditional point-to-point measurements can be taken, in addition to

volumetric measurements such as surface area and volume which are difficult to quantify on the

actual specimen (Tocheri, 2009). There have been several studies assessing the reliability of

taking measurements from a 3D model, but a comprehensive assessment using the standard

measurements used by forensic anthropologists has yet to be made. Park et al., (2006) compared

the intra- and inter-observer reliability between taking physical caliper measurements and virtual

measurements of the cranium. A total of thirty unidentified skulls were scanned using a handheld

laser scanner, which served as the sample for two examiners to obtain thirty-three measurements.

The conventional measurements were obtained by having one examiner conduct two sets of

measurements with traditional calipers within a one week interval, which also tested the intra-

observer reliability. The inter-observer reliability was tested by having each examiner collect the

virtual measurement sets twice, within a one week interval. The intra- and interobserver

reliabilities were tested using the intra-class correlation coefficient (ICC). The results showed

high ICC values, and the researchers concluded that the laser scanning and virtual measurement

technique would be able to replace the conventional measurement technique. However, the

measurements used in the study were arbitrary landmarks usually used for skull-tophoto

superimposition and facial reconstructions, in addition to seven random parameters used

frequently in craniometry. The lack of complete standard measurements used in this study

presents a serious flaw, despite the strength of the statistical methods employed. Tocheri (2009)

also contends that if traditional measurements can still be obtained from the cranium, then

digitization and virtual measurement are not justified. Nonetheless, the ability to take virtual

measurements using a 3D model of a cranium when the original is no longer available is highly

advantageous.

In contrast to performing the traditional point-to-point measurements on a 3D model is

the ability to take non-standard measurements such as surface-area and surface-curvatures, which

are difficult or impossible to quantify on the actual specimen (Tocheri, 2009; Sholts et al., 2010).

Sholts et al., (2010) suggest that volumetric measurements such as volume and surface area

which are afforded by the use of a 3D model can provide more precise determination of sex. This

can further aid in developing more accurate biological profiles, which can assist in making a

positive identification of unknown human remains. Sholts et al., (2010) performed a similar

study to Park et al., (2006) but assessed the accuracy and repeatability of measuring cranial

volume and surface area using 3D models that result from laser scan data. Testing the inter- and

intra-observer error of non-traditional measurements derived from laser scan data is necessary in

order to establish the reliability of this technique for use in forensic anthropology. In their study,

this was accomplished by having two operators perform three separate scans of five human

crania, using the NextEngine Desktop 3D scanner. The two operators utilized different protocols

for scanning and processing (see Sholts et al., 2010), and the resulting 3D models were compared

by measuring total volume and surface area using the RapidWorks 2.3.2 software by

NextEngine™ Inc. After the results were compared, the authors found that the differences

between the operator’s distinct protocols for scanning and processing did not result in a

significant statistical deviation in surface area or volume for the sample of skulls that were

scanned. The size of the mesh triangles however did affect the precision of the measurements.

The measurements were less accurate the greater the size of the mesh triangles, corresponding to

lower resolution 3D models.

In conclusion, the researchers found that the precision of the measurements recorded from

3D models which were created by two different protocols demonstrate the reliability of using

laser scanned models in acquiring metrical data such as volume and surface area. Considering

that at present there are no standard techniques for the measurement of cranial volume and

surface area, the authors recommend that scanning and processing protocols be documented

whenever such measurements are employed. Despite the positive conclusions of

Park et al., (2006) and Sholts et al., (2010), both studies are of questionable value in the field of

forensic anthropology because neither study was carried out using the standard set of cranial

measurements used by forensic anthropologists (Buikstra & Ubelaker, 1994). Because of this,

future research testing the reliability of taking measurements from laser scan data should

incorporate the standard measurements used in forensic anthropology.

Visualization and Illustration

After being processed and saved to hard-disk or other digital storage formats, the

proceeding application is the capability for visualizing the surfaces on the computer screen. This

includes the ability to rotate, zoom and pan around the digitized surface (Niven et al., 2009). A

practical way to view 3D models is through the use of 3D PDFs, which allow the user to rotate

the model, to zoom in and out for closer inspection, manipulate lighting, in addition to numerous

other functions. The use of 3D PDFs to visualize anatomical structures has been employed in the

visualization of molecular structures (Kumar et al., 2010) and for viewing faunal remains (Niven

et al., 2009), but could be easily adapted for the visualization of bone in forensic anthropology.

In addition to interactive PDFs, non-interactive video clips can also be generated for use in

court, academic presentations and education (Niven et al., 2009). A 3D model can be integrated

into

Microsoft® PowerPoint® presentations using 3D PDFs, and presented in a variety of situations.

In court, a model can be presented as evidence or as a visual aide when discussing identity and

trauma in lieu of potentially disturbing members of the court by presenting the actual remains

(Komar et al., 2012). The models can also be used in education (Davy-Jow et al., 2012), whether

in the classroom, anthropology teaching lab, or conference.

Virtual Reconstruction

Virtual reconstruction is the virtual reassembly of fragmented bone, which can then be

made physical again through rapid-prototyping (Zollikofer and Ponce de Leon, 2005). Although

this technique is typically used with CT scanning in paleoanthropological applications (Kalvin et

al., 1995; Zollikofer et al., 1995), it can also be implemented using laser scanned images in order

to reconstruct fragmented elements of the human skeleton (Benazzi et al., 2009; Fantini et al.,

2008; Kuzminsky & Gardiner 2012). As an alternative to the traditional practice of reassembling

bone fragments using adhesive materials such as glue, virtual reconstruction is advantageous

because it allows the evidence to remain in its original condition (Fantini et al., 2008). This

circumvents several complications that could incur with the traditional method of reassembly, for

instance, if a ‘missing piece’ was found at a later point in time or the remains needed to be

reexamined in a disarticulated state.

Trauma Reconstruction

Trauma induced by sharp or blunt weapons on the surface of bone can be documented

using 3D surface scanning and reconstructed in a virtual environment, as investigated by Thali et

al., (2003). This is done by digitizing both the weapon and location of injury, and then orienting

the 3D models within a computer graphics program to test the fit between injury and instrument

that may have caused it. From this, a basic animation can be created using software such as

Blender (open-source), Autodesk® Maya® or 3ds Max®. This animation can be used as a visual

aide in court. After being digitized, physical replicas of the wound area and injury-causing

instrument can also be created, and used similarly in court (Thali et al., 2003).

Forensic Facial Reconstruction

In tandem with virtual reconstruction is forensic facial reconstruction, a process which

results in a facial likeness of the unidentified individual. A photograph of the likeness is then

distributed through public media in hopes that someone will recognize the individual. This can

lead to a positive identification through fingerprints, dental records or DNA analysis. Intact skulls

can be prototyped for the forensic artist to perform a facial reconstruction, but fragmented skulls

should be reconstructed in a virtual environment before they are reproduced.

Unfortunately, many forensic anthropology labs and medical examiners offices cannot afford to

make 3D-prints of every unidentified victim, and facial reconstructions are sometimes performed

directly on the skull or on casts made through traditional mold-making and casting processes.

These techniques can alter the evidence, and can be potentially destructive.

An alternative method to the traditional techniques of forensic facial reconstruction in

clay is the possibility of performing the reconstruction on the computer utilizing 3D graphics

software. This includes two primary techniques: the use of automated facial approximation

software, or the utilization of 3D modeling and digital sculpting programs to perform manual

facial reconstructions. In reference to computerized facial reconstruction, Wilkinson (2005)

differentiates between two traditions, that of facial approximation and facial reconstruction.

Facial approximations deal with generalized facial ‘types’ developed from the basic

characteristics of the skull. These approximations can be performed automatically by specialized

computer techniques such as the first one developed by Vanezis et al., (1989) as part of the

Wolfson transputer-based workstation project at the University College of London. These

computer techniques often work by deforming a scan of a human face over the digitized skull,

and further refinements are made using parametric transformations and warping (Quatrehomme

et al., 1997; Jones, 2001). Interpolation of volumetric data is another technique utilized in

automated computerized facial approximations (Evison, 1996; Michael & Chen, 1996). The

primary limitation of these systems that Wilkinson (2005) describes is that the resulting

approximation will always have some resemblance to the original facial template. Additionally,

the database of facial templates used is not comprehensive, the method relies too heavily on

tissue depth data, and accuracy studies have yet to be performed (Wilkinson, 2005).

In contrast to facial approximation, more detailed facial reconstructions can be performed

using 3D modeling and animation programs, as well as digital sculpting software. Wilkinson

(2005) employs a virtual sculpting program called Freeform® from Sensable Technologies® that

uses haptic feedback which allows the user to ‘feel’ the surface of the skull, further enhancing the

analysis and reconstruction of the soft tissue. Unlike the automated systems, this method requires

3D modeling skills, anthropological and anatomical knowledge. A 3D model of the skull is also

required, and can be easily obtained through the use of a laser-scanner.

3D-printing & Rapid Prototyping

In conjunction with surface digitization, 3D printing is a powerful tool that allows for a

model of evidence to be created. In addition to its application in forensic cases, 3D printing can

also be used to develop teaching collections in an anthropology teaching lab, and for hands-on

museum exhibits for educating the public. Having a prototype model of the original specimen

also helps to prevent over handling of the original, and a mold can be made from the replica, with

further copies produced at a lower cost than 3D printing multiples (Fantini et al., 2008).

Summary

In summary, the literature was found to provide a considerable reference for exploring the

applications of laser scan data for use in forensic anthropology. Although there are numerous

applications to be examined, only a few have been chosen to be explored in this thesis due to

budgetary constraints, time and the limited expertise of the author. Moreover, the visual

applications of laser scan data in other fields such as cell biology (see Kumar et al., 2010) and

zoology (see Niven et al., 2009) can be adapted for use with human bone in forensic

anthropology. This includes the creation of 3D PDFs, which will be achieved in this thesis by

using Adobe® Photoshop® CS5 and Adobe® Acrobat® X Pro. Embedding a 3D model within a

PDF also enables the user to acquire digital measurements; the reliability of this technique will

be tested in Chapter Five. The 3D model that results from laser scan data can also be used to

create 2D illustrations, which will be created in this project through the use of Pixologic™’s

ZBrush® 4r4. At the suggestion of Steyer et al., (2010), this project will attempt to integrate the

use of laser scan data and ZBrush® 4r4, in order to reassemble and virtually reconstruct damaged

and missing bone fragments. In addition to the applications, one particular issue highlighted by

Sholts et al., (2010) is the lack of scanning and processing protocols which are necessary for

establishing laser scanning as a reliable and repeatable methodology for documenting case

materials in forensic anthropology. This challenge will be addressed through an exploratory effort

of documenting techniques for scanning different bone types. The procedures documented in this

project could potentially serve as a foundation for developing standard protocols for digitizing

human remains using a laser scanner. The documentation of scanning and processing protocols,

in addition to the exploration of the visual applications, will form the basis of the research in this

project.

CHAPTER THREE: METHODS

Introduction

The methods aspect of this research project encompass the laser scanning equipment and

scan data processing software, in addition to the 3D graphics programs which the resulting scan

data can be imported into. These methods can be divided into two primary workflows. The first

workflow is linear and involves scanning the object of interest, and then proceeding to process

the scan data to create a complete 3D model. The discussion of this workflow will include

strategies for positioning the bone using the part-gripper of the scanner, setting up scan settings

for particular bones, suggestions for automatic and manual alignment of scans and further

processing of the scans to create a complete model of the original object. The second workflow

is divided into several sets, each describing a specific application for the scan data that can be

achieved using different computer graphics programs.

Adobe® Photoshop® CS5, is a 2D graphics editing program popularly used for photo

editing and digital painting, although it has several features relevant to 3D. In this project it is

used in the production of 3D PDFs, allowing the 3D model in OBJ format to be converted to a

U3D file. Adobe® Acrobat® X Pro, which is utilized in the creation of professional quality PDF

files, allows the U3D file to be embedded into a PDF as an interactive 3D model. ZBrush® 4r4

by Pixologic™ is a powerful digital sculpting program used in the film and video game industry

for the creation of high-resolution 3D assets. Here it is employed in the creation of rendered still

images of the scan data, and in the virtual assembly and reconstruction of fragmented bone.

Sample

A primary consideration in digitizing skeletal anatomy is the type of bone being scanned.

The morphology of human bone varies significantly throughout the body, and the complex forms

of certain bones such as those of the cranium can pose a challenge when attempting to acquire a

complete representation of them within a 3D model. In order to address this issue, an assortment

of bones was scanned to provide a research sample. This allowed specific complications

regarding the digitization of skeletal anatomy to be discussed. The specific bones scanned

represent the breadth of morphological diversity in the human skeleton and include the vertebrae,

scapula, clavicle, os coxae, sternum, rib, femur, cranium and mandible. The samples used in this

project were commercially prepared teaching examples of actual human bone which were

provided by the Anthropology Teaching Laboratory at the University of Central Florida.

3D Scanner Hardware and Software

The laser scanner chosen for this research project is the NextEngine™ Desktop 3D

Scanner (Figure 1). It was chosen due to its relative affordability (Kuzminsky & Gardiner, 2012),

resolution and mobility (Slizewski & Semal, 2009). It is a laser triangulation type of 3D scanner,

which uses a built in digital camera to identify the distance and angle of the laser stripes

projected from the laser emitter. The scanner also includes an automated turntable known as the

AutoDrive, which can be manually and automatically rotated within the ScanStudio HD Pro™

software. Using the automatic rotation feature, the scans are automatically aligned within the

software. When a rotation scan does not result in complete coverage of the object, the manual

rotation settings allow precise positioning of the object in order to obtain missing data.

Figure 1: The NextEngine desktop laser scanner (left) with AutoDrive and part-gripper (right). The part gripper has

been modified with an additional supporting prong (A).

The NextEngine™ Desktop 3D Scanner comes with a proprietary software package

known as ScanStudio HD™ which is used with the scanner hardware to process the scan data.

An upgraded version of the software, ScanStudio HD Pro™ was purchased for this project and

has the benefits of doubled scan speed, four times the raw point output (resulting in more detailed

capture), and has an additional large area format scan mode that increases the range of the

scanner (http://www.nextengine.com/products). ScanStudio HD™ is necessary for operating the

NextEngine Desktop 3D Scanner, but once the initial scans have been acquired, the data can be

processed in other scan data processing software such as MeshLab or Geomagic Studio™.

Figure 2: The ScanStudio HD Pro™ interface displayed after opening the program. The Scan button (A) used to

initiate the scanning process is circled in red.

Laser Scanner Process

Scanning the bones is the first step in exploring the applications of a laser scanner in

forensic anthropology. Unfortunately, it is not a simple task and there are many issues that need

to be considered in order to optimize the breadth and quality of the 3D data that is being

acquired. The general process for scanning and processing the scan data of human bone will be

outlined here in the methods chapter. The specific problems regarding the digitization of different

bone types will be elaborated in the results chapter. After the hardware and software has been

properly installed, the scanner hardware is activated by opening the ScanStudio HD Pro™

software. The scanning process is initiated by clicking the Scan button (Figure 2 A) in the top

menu, which opens a dialogue page where the scan settings can be specified. Displayed on this

page in addition to the scan settings is the viewport (Figure 3), where the object intended for

scanning is seen by the camera of the scanner. This window is used as a reference for manually

positioning the object for optimal scanning, in conjunction with changing the distance of the part

gripper from the scanner. A region of interest can also be selected by clicking and dragging

within this viewport, which will cause the scanner to scan and display the area that is selected.

Figure 3: The scan settings dialogue page within ScanStudio HD Pro™. The settings are on the left, and the

scanner’s camera viewport is on the right.

Positioning

The first parameter of the scan settings that can be set is the positioning mode (Figure 4),

which can either be a 360 degree full rotation of the turntable, a three scan bracket mode, and a

single scan mode which does not result in movement of the turntable during scanning. The 360

scan mode utilizes the AutoDrive settings and should be used whenever possible in order to

streamline the scanning process by using the automatic alignment feature of the AutoDrive. This

setting results in an automatically aligned scan set based on the divisions input by the user. It

does not result in a complete model, since the part-gripper base and support prongs prevent the

laser stripes from reaching certain areas. Nevertheless it results in greater coverage and efficiency

than the other scan settings. A bracket scan is a series of three scans that can be used to capture a

wider range of missing data than the single scan setting. This is useful in providing a greater

range of anatomical features and can be used to align the bracket scan with other scan sets. The

bracket setting is also more efficient than the single scan setting because the three scans will be

automatically aligned by the ScanStudio HD Pro™ software. Single scans can be necessary when

trying to capture hard to reach areas or if the missing data does not require more than one scan.

Single scans require manual alignment, whether they are being aligned to 360 scan sets, bracket

sets or other single scans. Single scans can present problems if there are not enough anatomical

features to be used for manual alignment.

Figure 4: The positioning and divisions section of the scan settings dialogue page within ScanStudio HD Pro™.

Divisions

In order to change the degree increments at which the turntable rotates during the 360 and

bracket position settings, it is necessary to change the ‘Divisions’ section of the scan dialogue

page. The amount of divisions equals the amount of times the turntable rotates within 360

degrees. To calculate the incremental degrees of rotation, divide 360 by the amount of divisions.

The maximum number of divisions is 16, which allows the minimum degree of incremental

rotation: 22.5 degrees. A division of 8, at 45 degree incremental rotations was found to provide

sufficient coverage and overlap between scans which proved useful in aligning the scans of bone.

It was also performed in a reasonable amount of time, and without an excess of overlapping and

redundant data.

Resolution (Points / In.²)

The resolution of the scans is determined by manipulating the Points / In² slider (Figure

5). The slider is divided into three sections: Quick, SD and HD. Each of these sections has three

tick marks which increase the amount of surface points acquired by the scanner, from left to

right. Moving from left to right under each of the sections will result in the Time slider at the

bottom of the screen to increase, and moving from each tick mark results in an increasing

percentage of memory (RAM) to be used. In order to scan the sample of this research within a

reasonable amount of time, the highest points per square inch of the SD setting (10,000) was

chosen as a compromise between scan acquisition time and resolution. Because of the many

issues encountered during the scanning of different bone types, the SD settings were utilized for

efficient time-management in troubleshooting and establishing protocols for scanning. With the

HD settings, the scanning process would have taken three times longer, therefore increasing the

amount of time it would have taken to resolve the issues that were encountered. Once these

scanning protocols have been established and implemented, the HD settings should be utilized in

order to obtain the most accurate representations of the objects being scanned. An 8 division 360

scan took 10.3 minutes with this setting, as opposed to the 27 minutes it would take to scan using

the HD settings. The amount of time required for a particular type of scan is displayed on the

Time slider at the bottom of the scan settings page.

Figure 5: The Points / In.² and Target section of the scan settings dialogue page within ScanStudio HD Pro™.

Target

This section relates to the lightness of the object being scanned. The default neutral

setting was maintained throughout the scanning process of this thesis (Figure 6). The

NextEngine™ user manual does not discuss this feature in depth, but it is assumed that it

modifies the laser intensity for scanning light and dark objects.

Figure 6: The Time and Memory section of the scan settings dialogue page within ScanStudio HD Pro™. It is found

at the bottom of the screen, and has a direct relationship to the Points / In.² slider.

Range

There are three range settings which determine the field of view of the scanner (Figure 7):

Macro, Wide and Extended. The macro setting results in the highest resolution scans, but has a

smaller range which requires the object to be moved closer to the scanner, between 12.7 cm and

22.86 cm (5” and 9”). This in turn can require several more scans if the entire object does not fit

entirely in the scan preview window. The wide setting extends the range between 38.1 cm and

55.88 cm (15” and 22”), but results in an overall lower resolution scan. The extended setting

further increases the range of the scanner between 38.1 cm and 76.2 cm (15” and 30”) and results

in a considerably lower resolution that is not practical for digitizing bone.

Figure 7: The Range section of the scan settings dialogue page within ScanStudio HD Pro™.

Texture

The NextEngine™ scanner captures texture information, which is data that contains the

color of an object in computer graphics terminology. Texture information may be of only

secondary importance if the surface of the object is the focus of the scanning project, but it can

aid in highlighting small anatomical features (Slizewski et al., 2010; Niven et al., 2009). The

texture settings are not determined in the scan settings dialogue box but should be considered

prior to scanning. The visualization of textures can be temporarily disabled by clicking the

‘Shaded’ button in the bottom right corner of the interface. Textures can be prevented from

loading whenever a project is opened by going to the Edit tab, to Preferences and clicking on the

No Textures bubble (Figure 9). This does not disable the scanner from acquiring textures, but

merely prevents them from being displayed in the program and is a way of increasing computer

performance when working with large scan data sets. The visualization of textures can be

reenabled by going back to the Preferences and clicking on the 2D texture display bubbles. The

scanning in this project was performed without the visualization of texture within the program to

increase computer performance for faster scanning and processing. The texture information

captured by the NextEngine™ scanner does not always yield quality results (Figure 8). If

highquality textures are desired it is recommend that they are acquired through other means, such

as through photo-texturing using a 3D modeling program like Autodesk® Mudbox® or ZBrush

4r4 by Pixologic™.

Figure 8: A final model of the clavicle created using the NextEngine™ desktop laser scanner and ScanStudio HD

Pro™ with textures enabled.

Figure 9: The ScanStudio HD Pro™ preferences dialogue box, which can be accessed at the bottom of the Edit

drop-down menu of the main toolbar. The options for displaying textures are seen at the bottom of the window.

Scan Data Processing

Once the object has been scanned it is necessary to process the scan data effectively so

that the finished 3D model is as complete and accurate as possible. This process is outlined in the

Next Engine™ user manual, but there are different techniques that can be implemented

depending on the intended use of the scan data. The specific procedures developed for processing

scan data of different types of bone will be documented and outlined within this project (Figure

10).

Figure 10: A flow-chart illustrating the basic work-flow for scanning and processing the scan data.

Trimming

Once the scans have been obtained it is important to trim and remove the turntable,

partgripper components and any other irrelevant data from the scan data sets before alignment

(Figure 11 A). This should be performed with all scan data sets prior to alignment. This is done

by clicking on the Trim button on the toolbar of the ScanStudio HD Pro™ software. It is then

possible to utilize different selection tools to highlight the unwanted data (Figure 11 B), which

will be displayed in red (Figure 11 C). Once the unwanted data has been selected, the user simply

clicks the Trim button and the software will remove the data from the scans (Figure 11

D). Further removal of overlapping data can be performed, but should only be done after

alignment in order to maintain a sufficient amount of anatomical features which can be used as

reference points during alignment.

Figure 11: This image illustrates the process of trimming excess data from a 360 scan set of a clavicle. The part

gripper (A) should be trimmed from the scan data prior to alignment. Precise selections can be made using the

polygon selection tool (B). Selected areas are displayed in red (C), and are removed after clicking on the trim button

(D).

Alignment

Aligning scans is a necessary step when acquiring a complete model of the original

object. The automatic alignment features are efficient, but since all the surface data of an object

cannot be completed with a single 360 scan set, manual alignment is still necessary. Manual

alignment is performed by clicking on the Align button on the main toolbar, which opens up the

split-view alignment screen (Figure 12). The software then prompts the user to place three or

more pins to align the scans. This is done by dragging three different colored alignment pins onto

different areas of the two scan data sets that are displayed. The same colored dots should be

placed onto the same areas between the scan sets, so it is necessary for there to be some amount

of overlapping data between the scans. At least three matching points are required in order to

align the scans together (Figure 12 a, b, and c) but more than three can be placed in order to

increase the accuracy of the alignment. Anatomical features and unique patterns on the bone

serve as a great reference for aligning scan data together. Occasionally, the align button will not

respond even after placing the three alignment pins. This is because the pins were not placed

accurately enough between the scan data sets. When this happens, the user can either reorient the

pins into a more accurate position or click the refine button, which will automatically refine the

placement of the pins. After clicking align, the scan sets should appear aligned without any

visible errors (Figure 13).

Figure 12 A scan of the clavicle is used to illustrate how the alignment pins are placed on analogous anatomical

features between the scan data sets. The first set of necessary alignment pins are red (a), blue (b) and yellow (c).

Figure 13: After placing the alignment pins and clicking the Align button, the two scan sets of the clavicle will be

aligned together.

Fuse

Once all the necessary data has been obtained, trimmed and aligned, the scans can be

combined to create a complete model using the fuse function, which is accessed in the main

toolbar. The resulting model can be simplified, which decreases file size and resolution, by

adjusting the tolerance of the simplify slider found on the fuse mode toolbar. Further attributes of

the fuse mode can be modified by clicking on the settings dial, which opens up a new window

(Figure 14). In this window it is possible to adjust the hole-filling settings, the texture blending

and resolution ratio. Under the hole-filling settings it is possible to create a water-tight model,

which is a solid model without any holes in the surface. A water-tight model (Figure 15) is

necessary for outputting the model for 3D printing and rapid-prototyping, and is also desirable

when working with 3D modeling and animation software. Alternatively, the user can modify the

hole filling settings manually, or choose the setting for no hole-filling. Unlike traditional 3D

models which are created by a designer on the computer, scan data generally contains open areas

in the surface (holes) which can complicate advanced post-processing of the data for

visualization and rapid-prototyping. Texture blending modifies the blending of the textures to

account for brightness variations, and the resolution ratio determines the size of the vertices of

the resulting mesh in relation to the model prior to fusing scans. Once all the settings have been

determined, the function is completed by clicking on the fuse button.

Figure 14: The fuse settings dialog box within ScanStudio HD Pro™.

Figure 15: After fusing the scans of the clavicle together, any overlapping data is removed. A water-tight model

without any holes can be created, which is ideal for advanced post-processing for visualization and

rapidprototyping.

Export

Once the scan data has been processed in the ScanStudio HD Pro™ software, the

resulting 3D model should be exported by going to the main toolbar, clicking File, Save As and

changing the Save as type to the OBJ file (*.obj) format (Figure 16), which is a standard 3D

model file format that can be imported into most 3D graphics software. Along with the OBJ file

which contains the 3D information, an MTL file and a series of JPG images are also saved,

which contain the material and texture information created by the scanner.

Figure 16: The Save As dialog box within ScanStudio HD Pro™ is used for exporting 3D files, which can then be

imported into other 3D modeling programs.

Computer Graphics Software

Once a 3D model has been exported from the ScanStudio HD Pro™ software, it can be

imported into a variety of 3D graphics programs and used to create visual tools that can aide in

forensic investigations. Three computer graphics programs will be used to explore the

applications for using the laser scanned data in this project: Adobe® Photoshop® CS5, Adobe®

Acrobat® X Pro and ZBrush™ 4r4 by Pixologic™. Adobe® Photoshop® CS5 is a standard 2D

photo editing and painting program that has limited 3D capabilities. It is used in conjunction with

Adobe® Acrobat® X Pro to create 3D PDFs, and can be used to edit and annotate rendered

images from ZBrush 4r4. Adobe® Acrobat® X Pro is used for creating professional-level PDF

files that have audio, video and 3D visualization capabilities. In this project it will be used

specifically for its ability to embed interactive 3D models into a PDF. Pixologic’s ZBrush 4r4 is a

powerful digital sculpting program used in the professional art production workflows in the

gaming, toy and movie industry. It is particularly appealing in this project due to its ability to

handle millions of polygons, allowing it to handle high-resolution scan data with ease. In

conjunction with this ability, it will be used to explore virtual reassembly and reconstruction

techniques.

CHAPTER FOUR: RESULTS - SCANNING

Introduction

The unique morphology of different bones types within the human skeleton requires

diverse strategies to be implemented for efficient scan data collection. This includes positioning

the bones with the part gripper in a variety of ways that will enable all of the surfaces to be

documented, modifying the position of the scanner to enable it to scan larger bones that are

beyond the ideal object placement range of the Macro setting, in addition to other methods

exclusive to particular bone types. Due to a lack of protocols in forensic anthropology for laser

scanning human bone, it is necessary to document scanning techniques for different bone types in

hopes of establishing the reliability of this technique. This chapter will provide the protocols that

were developed in this project, and may be used as a reference for future research and

improvement.

Laser Scanning Different Bone Types

Vertebrae

The majority of the surfaces of the vertebra can be obtained by employing two 360° scan

sets with eight divisions each, one scan set made with the vertebral body placed horizontally

(Figure 17 A), and the other with the vertebral body placed vertically (Figure 17 B) on the part

gripper base. Figure 18 A & B show the results of these scans. These scans were then aligned

using the superior articular facets (Figure 19 a & b) and spinous process (Figure 19 c) as

references for alignment. The most challenging aspect of scanning the vertebra is in obtaining the

surfaces which border the interior of the vertebral foramen, the inferior articular facets and the

posterior surface of the vertebral body (Figure 20). In an attempt to acquire these surfaces,

several single scans were collected with the vertebra oriented with these areas facing the scanner.

Some of the missing areas were obtained, including the posterior surface of the vertebral body

(Figure 20 a; Figure 21 a). Although several attempts were made, the surface contralateral to the

posterior surface of the vertebral body was not obtained (Figure 21 a). The Fuse function was

attempted despite the missing data (Fig. 21 a), which caused the software to completely fill in the

vertebral foramen (Fig. 21b). The ScanStudio HD Pro™ hole-filling feature of the Fuse function

is a global operation and does not allow the holes to be manually selected and filled. In this

instance, it would be useful to utilize additional software to overcome the limitations of

ScanStudio HD Pro™.

Figure 17: The initial scans of the L5 vertebra consisted of two 360° scan sets, one with the long-axis of the bone

oriented horizontally (A) and one with the long-axis of the bone oriented vertically (B).

Figure 18: The horizontal (A) and vertical (B) 360° scan sets of the L5 vertebra within the ScanStudio HD Pro™

software, prior to alignment.

Figure 19: The superior articular facets (a, c) and spinous processes (b) of the L5 vertebra were used for aligning the

360° scan sets.

Figure 20: The horizontal and vertical 360° scan sets of the L5 vertebra are aligned (A), but the interior borders of

the vertebral foramen are still missing (a). Several more scans were collected (B), and the posterior border of the

vertebral body was obtained (b).

Figure 21: The posterior border of the vertebral foramen, intermediate to the superior and inferior articular facets

was not obtainable (A, a). Despite this, the scan data of the L5 vertebra was processed with the Fuse function. The

missing data caused the software to bridge the empty spaces, filling the vertebral foramen (B, b).

Scapula

First, a 360° scan set was collected with the superior border and acromion process placed

on the surface of the supporting stand of the turntable (Figure 22 A). A second 360° scan set was

created with the breadth of the scapula placed horizontally was created in order to obtain the

anatomical borders of the bone (Figure 22 B& Figure 23). This secured the bone from moving

during the scanning process, and allowed the flat surfaces of the subscapular and infraspinous

fossa to be obtained. Once these scan sets were completed and examined in ScanStudio HD

Pro™, the thin surfaces of the fossae caused an issue with the 360° scan set. In observing the

scan data, the fossae did not have the same shading as the other surfaces, which have a lighter

blue color (Figure 24). This difference in color is how the software illustrates different surface

normals. In computer graphics, surface normals govern a surfaces orientation in relation to a light

source during shading. The ScanStudio HD Pro™ displays outer surfaces as a lighter blue color,

which represents the outer surface of the object within the viewport. The darker blue shading

represents the inner surface of the object and is useful for identifying missing data where a

surface still needs to be acquired. The surfaces of the fossae presented a dark blue color,

indicating that an internal surface was protruding through the external border of another surface.

After expanding the 360° scan family and detaching individual scans, it was found that the

inversion of the surface normals occurred when scans of the two different fossae were aligned

together. The inversion of the surface normals was caused by an intersection of the thin surfaces

of the fossae, which may not have been distinguishable by the laser triangulation system. After

deleting the problem scans, the 360° scan sets were aligned together, and resulted in the same

problem.

A second scapula that appeared to be slightly thicker than the first (Figure 25 A& B) was

scanned using the same protocols. The first 360° scan set exhibited similar surface intersections

as the first scapula, but appeared unevenly on the surface of the fossae (Figure 26 A). A second

360° scan set did not appear to have any surface intersections (Figure 26 B). A third 360° scan set

was collected with the scapula lying flat on the part-gripper base (Figure 26 C). Several attempts

were made to align these scan sets together (Figure 27), but the attempts resulted in errors each

time (Figure 28). This occurred despite attempting to align the sets in different orders, or using

different anatomical features as a reference for setting the alignment pins. Because of the

problem with the translucency, a complete model of the scapula was unable to be obtained.

Figure 22: For the initial 360° scan set the scapula was placed with the long-axis oriented vertically with the

superior aspect placed on the base of the part-gripper (A). The supporting prong was placed against the infraglenoid

tubercle to prevent the bone from moving during the rotation of the turntable. An additional 360° scan was collected

with the scapula placed flat on the base (B) in order to scan the anatomical borders of the bone.

Figure 23: A 360° scan set with the scapula placed horizontally on the part-gripper base was completed in order to

acquire the surfaces of the anatomical borders.

Figure 24: The 360° scan set with the scapula oriented vertically on the part-gripper base presented a problem. The

thin surfaces of the subscapular and infraspinous fossae were penetrated by the lasers of the scanner. This caused the

scanned surfaces to intersect, creating the dark-blue inverted surface normals in the section outlined by the red

ellipse.

Figure 25: A side by side comparison of the first (A) and second (B) scapulae that were scanned. The fossae of the

second scapula appear to be slightly thicker and the bone is also of a darker color.

Figure 26: The second scapula was scanned using the same protocols as the first. The first 360° scan set (A)

exhibited similar intersections to the first scapula, whereas the second (B) and third (C) 360° scan sets did not.

Figure 27: Analogous regions of the scapula (a, b & c) were chosen between the scans for the placement of the

alignment pins.

Figure 28: Despite several attempts at aligning the scan sets of the scapula by using different regions as a reference

for the alignment pins, the alignment process consistently resulted in errors.

Clavicle

The first 360° scan set was performed with the sternal end of the clavicle placed on the

part gripper base, and the supporting prongs of the part-gripper positioned in order to pinch the

acromial end and conoid tubercle, supporting the top (Figure 29 A). The height of the base was

oriented in order to scan as much surface area as possible, but the full length of the clavicle was

not obtainable in a single scan with the recommended distance of 6.5” for the Macro settings

(Figure 30 A). The base was lowered in order to scan the acromial end of the clavicle in a

separate 360° scan set (Figure 30 B). The vertical scans captured much of the surfaces, but did

not acquire the sternal and acromial ends (Figure 31 A&B). A few single scans were attempted in

order to capture the missing data, but did not supply enough features to permit an accurate

alignment with the initial 360° scan sets. Further 360° scan sets were collected in order to obtain

sufficient anatomical features for ease of alignment. This was done by placing the clavicle flat,

with the long-axis parallel with the part gripper base, which resulted in the scans within Figure

32 (A &B). One scan was performed with the superior surface of the clavicle placed flat on the

part-gripper base (Figure 29 B). Another scan set was created with the inferior surface of the

clavicle placed flat on the part-gripper base (Figure 29 C). The scan sets were then aligned by

referencing the anatomy: the muscle attachment of the deltoid on the anterolateral surface (Figure

33 A), the conoid tubercle (Figure 33 B), and a nutrient foramen that was located medial to the

trapezius muscle attachment (Figure 33 C). After aligning the scans together, they were then

fused (Figure 34). The process of scanning the clavicle was straightforward and required minimal

alignment, presenting no exceptional challenges.

Figure 29:The surfaces of the clavicle were obtained through a series of three 360° scan sets: one with acromial end

place on the part-gripper base (A), and two more with the bone placed lengthwise, one with the inferior surface

upward (B) and another with the superior face upward (C).

Figure 30: Two 360° scan sets were acquired with the clavicle oriented vertically. The first (A) scan set acquired the

majority of the surfaces, except for the acromial end. The part gripper was then lowered so that the surfaces of the

acromial end of the clavicle (B) could be collected.

Figure 31: The surfaces of the sternal (A) and acromial (B) ends of the clavicle were missing from the 360° scan sets

where the bone was oriented in a vertical position.

Figure 32: The acromial (A) and sternal (B) ends were acquired by orienting the clavicle horizontally and placing it

flat on the part gripper base.

Figure 33: The muscle attachment of the deltoid on the anterolateral surface (A) and the conoid tubercle (B) were

used as references for alignment of the vertical scan sets of the clavicle, in addition to a nutrient foramen that was

located medial to the trapezius muscle attachment (C).

Figure 34: A view of the inferior surface of the final 3D model of the clavicle once the scan sets have been aligned

and fused together.

Os Coxae

The first 360° scan set was created with the ischiopubic ramus placed flat on the part

gripper base (Figure 35 A). An extra adjustable horizontal post was taken from another

partgripper and combined with the one mounted on the scanner. The iliac blade was pinched

between the rubber tips of the adjustable horizontal posts, enabling greater support to the bone as

it rotates during the automated scanning process. This initial 360° scan set allowed much of the

surfaces to be scanned, but the surfaces of the iliac crests, ischial tuberosity and ischiopubic

ramus required additional scans. The outer rim of the ascetabular fossa was placed on the part

gripper base and another 360° scan was performed (Figure 35 B). Additionally, several more

single scans were acquired in order to obtain several areas of missing data that could not be

obtained using the automated scanning. Primary features used for alignment included the

acetabular fossa (Figure 36 a), the anterior inferior iliac spine (Figure 36 b) and the sciatic notch

(Figure 36 c). The particular specimen that was scanned had a few damaged areas where the

cancellous bone was exposed. This included the ischial spine (Figure 37 a), the tip of the

posterior inferior iliac spine (Figure 37 b) and the medial edge of the ischial tuberosity (Figure 37

c). The porous surface of the cancellous bone exhibited a problem for the scanner, and these

areas were unable to be scanned, leaving holes in these areas. The scan sets were fused with the

water-tight model setting and the holes were filled over without presenting any particular

problems (Figure 38 A& B).

Figure 35: For the first 360° scan set the long-axis of the os coxa was oriented vertically, with the ischial tuberosity

placed on the part-gripper base, and the iliac blade pinched with the supporting prongs (A). The second 360° scan

set was collected with the long-axis oriented horizontally, with the posterior aspect of the ascetabulum resting on the

part-gripper base. A single supporting prong of the part-gripper base was used to press down on the auricular surface

and secure the bone (B).

Figure 36: Numerous alignment pins were used to align the scan sets of the os coxa. Primary features used for

alignment were the acetabular fossa (a), anterior inferior iliac spine (b) and the sciatic notch (c).

Figure 37: The tip of the posterior inferior iliac spine (a), the ischial spine (b), and the medial edge of the ischial

tuberosity (c) presented cancellous bone on the os coxa that was difficult to capture with the laser scanner.

Figure 38: The lateral (A) and medial (B) views of the final scanned model of the os coxa after fusing the scans.

Sternum

The initial scan of the sternum was collected by placing the bone upright on the part

gripper base and pinching it between the adjustable horizontal support posts. Unfortunately,

during this initial scan, the bone was not pinched securely enough and moved slightly during the

scanning process. This caused a visible error during alignment, displaying two copies of the bone

within the same scan. In order to address this, the part gripper was further adjusted to secure the

bone properly. Once this was done, the surfaces of the sternum were obtained with two 360° scan

sets, one with the sternum upright in anatomical position (Figure 39 A), and another with it lying

flat on the part gripper base (Figure 39 B). These protocols resulted in the scan data shown in

Figure 40 (A & B). The scan sets were then aligned easily aligned using the costal notch of the

2nd left rib (Figure 41 a), the lateral right edge of the jugular notch (Figure 41 b) and the superior

edge of the costal notch of the 4th right rib (Figure 41 c). After alignment, the sternum scans

were fused (Figure 42A&B) requiring no further scanning.

Figure 39: The sternum oriented vertically (A) and horizontally (B) in two different 360° scan sets.

Figure 40: The horizontal 360° scan set (A) is missing the surfaces of the jugular and clavicular notches of the

sternum. The vertical scan (B) is missing some of the lateral edges of the corpus sterni.

Figure 41: The costal notch that articulates with the 2nd left rib (a), the lateral right edge of the jugular notch (b), and

the superior edge of the costal notch of the 4th right rib (c) were used as references to place the alignment pins.

Figure 42: An anterolateral (A) and posterolateral (B) view of the final model of the sternum after the scans were

aligned and fused.

Rib

Ribs are particularly challenging to scan because there are few substantial surfaces to

provide support when placing the bone on the part gripper base. Because of this, it took a

considerable amount of time to properly secure the bone with the supporting prongs so that the

bone did not move during the scanning process. The rib that was chosen to be scanned was a left

second rib. The first 360° scan was collected with the rib placed in anatomical position, flat on

the part gripper base (Figure 43). Two more 360° scan sets were possible after considerable

adjustment of the supporting prongs of the part gripper. One of these sets was acquired with the

sternal end placed on the part gripper base, and the vertebral end supported above by pinching

with the part-gripper prongs (Figure 44 A). The other set was created with the bone in the

opposite orientation, with the vertebral end lying on the part-gripper base and the sternal end

oriented superiorly (Figure 44 B). The three 360 scan sets resulted in the scan data shown in

Figure 45 (A, B & C). The scan sets were then aligned by referencing the tubercle (Figure 46 a),

and two areas where the bone had been slightly damaged (Figure 46 b & c). After alignment, the

scan sets were fused, creating a water-tight model (Figure 47A & B). Except for the difficulty in

securing the rib with the bone oriented length-wise in a vertical position, the rib was not

complicated to scan and process.

Figure 43:

The first 360° scan set was acquired with the length of the rib oriented horizontally.

Figure 44: The rib is placed in a vertical orientation for scanning. One set is collected with the sternal end resting on

the part-gripper base, and the other scan set was acquired with the vertebral end resting on the base.

Figure 45: The results of the horizontal (A) and two vertical (B, C) 360° scan sets of the rib.

Figure 46: The tubercle (a) was the only anatomical feature that was used for alignment between these scan sets of

the rib. The other alignment pins (b, c) were placed on small areas where the bone had been chipped.

Figure 47: Superior (A) and inferior (B) views of the final rib model.

Femur

The primary challenge in scanning the femur has to do with its length. Scanning within

the ideal range restrictions for the ‘macro’ settings allows only a portion of the bone to be

captured with the 360° scan mode. The first 360° scan set was collected with the femur oriented

in anatomical position, with the epicondyles lying on the part gripper base (Figure 48). The two

supporting prongs of the part gripper were adjusted to pinch the top of the femur in order to

stabilize the bone and prevent it from moving during scanning. The part gripper base was then

lowered so that the bottom of the bone was visible in the scanning window. The second 360°

scan set was acquired with the bone in the same position, but with the base lowered all the way

down to allow further coverage of the bone during scanning. The third set was created with the

femur flipped, with the head of the femur lying on the part gripper base. The supporting prongs

pinched the shaft of the femur just below the epicondyles. A fourth scan set was acquired

similarly as the second, with the part gripper base lowered as much as possible to increase

coverage. When attempting to align the first and second sets with the third and fourth sets, a

significant portion of the mid shaft of the femur was still missing and it was not possible to align

the sets together. The scan settings were changed from ‘macro’ to ‘wide’, changing the ideal

distance of the object from the scanner from 6.5” to 17”. This allowed a greater extent of the

femur to be scanned. However, after attempting to align the scan set collected with the ‘wide’

setting, the mid-shaft of the femur was found to contain few anatomical features that could be

used for alignment. The results of the wide scan settings also resulted in comparatively lower

resolution than the macro scan settings (Figure 50), compounding the challenge of manual

alignment.

A new strategy was attempted, by placing the scanner on a platform in order to increase

the vertical range of the scanner setup. Several large textbooks were stacked on top of each other

to create a platform that measured approximately 10” tall (Figure 48). This allowed the entire

length of the femur to be scanned in several 360° scan sets by adjusting the height of the

partgripper accordingly. Five 360° scan sets were acquired, each time adjusting the height of the

part gripper slightly to maintain overlap while capturing new data. The part gripper components

were not trimmed before aligning the scans since they provided references for alignment along

the mid-shaft where there are few distinct anatomical features to be used for matching the

alignment pins. After aligning the five scans, the part gripper was then removed with the trim

tool. The part gripper prongs prevented the laser stripe from reaching certain areas of the mid-

shaft, so an additional 360° scan set was created after adjusting the prongs to an area that in

which scan data had been obtained. After these 360° scan sets were created, two single scans

were acquired in order to capture the superior aspect of the head of the femur and the inferior

surface of the epicondyles (Figure 49). These were easy to align to the existing sets. After the

scan sets were aligned, a water-tight model of the femur was successfully created (Figure 51 A &

B). Due to the size of the femur, and the relatively featureless surfaces of the mid-shaft, the

femur was a challenge to scan, yet it was still possible generate an accurate water-tight model.

Figure 48: The first series of scans was created with the femur oriented in anatomical position. The height of the

part-gripper was then adjusted several times until the majority of the surfaces were obtained.

Figure 49: In order to obtain the inferior surfaces of the epicondyles and the superior aspects of the head of the

femur and greater trochanter, the length of the femur was oriented horizontally. A series of single scans was

collected, and aligned with the existing scan data.

Figure 50: A view of the posterior surface of the proximal end of the femur illustrates the difference in resolution

between the macro (A) and wide-area (B) scan settings.

Figure 51: An anterior (A) and posterior (B) view of the final fused model of the femur.

Cranium

The complex morphology of the cranium makes it the most challenging bone in the

human skeleton to document with a laser scanner. As with the other bones, the majority of the

surfaces were obtained with a series of 360° scan sets. The first scan set was created with the

cranium in anatomical position (Figure 52 A), and the second was created with the cranium lying

on its side (Figure 52 B). Bracket scans were collected with the cranium oriented in a variety of

positions depending on the missing data that was being obtained (Figure 52 C). These 360 scan

sets resulted in the scan data shown in Figure 53 (A, B & C). The underside of the cranium and

orbits were the most difficult areas to scan. For the underside of the cranium the most

challenging areas were the teeth, the posterior aperture of the nasal cavity and the surfaces of the

maxilla, in addition to the surfaces of the temporal and sphenoid bones that are medial to the

zygomatic arch. These surfaces were obtained using several bracket scans with careful

positioning of the cranium with the part-gripper.

Because the cranium has numerous openings and cavities many of which could not be

entirely scanned, the creation of a water-tight model could not be achieved without significant

distortions (Figure 54 A:a & b). In order to create a more accurate model without any holes, it

would be necessary to use a scan data processing program that has the ability to manually fill

individual holes in the scan data. An alternative model was created with the Fuse function

without the hole-fill setting (Figure 54 B).

Figure 52: Many of the surfaces of the cranium were obtained through a series of 360° scan sets. The first scan was

created with the cranium oriented in anatomical position (A). The second set was made with the lateral aspect placed

flat on the part gripper base (B). Multiple bracket scans were performed with the cranium oriented in different

positions depending on the data that was missing (C).

Figure 53: The greater part of the cranium was obtained through 360° scan sets with the bone oriented in a

horizontal (A) and vertical (B) position. Bracket scans were used to obtain missing data (C).

Figure 54: Fusing with the water-tight settings (A) produced distortions in the left orbit (a) and the nasal cavity (b).

These distortions were to significant to ignore, so the Fuse function was performed without an hole-filling settings

(B).

Mandible

As with the other bones, the mandible was digitized through a series of 360 scan sets. The

first scan set was collected with the mandible in anatomical position (Figure 55 A), with the

inferior border of the horizontal ramus resting on the part-gripper base. The second and third sets

were collected with the lateral surface of the ascending ramus placed flat on the part-gripper

base, one with the right ascending ramus (Figure 55 B) and the other with the left ascending

ramus placed on the part-gripper base (Figure 55 C). After aligning these scan sets together, the

majority of the data was captured. However, the scan data of the teeth did not present a clean and

accurate surface. The scanning of the teeth resulted in surface noise and artifacts between the

interproximal surfaces of the teeth (Figure 56 a & b). This is likely caused by the slight

translucency of the enamel, which would absorb the lasers coming from the scanner. The

problem areas were removed by detaching the affected scans from the original scan sets and

manually removing the undesired areas using the polyline selection tool (Figure 57: A, B, C

&D). A water-tight model was created using the Fuse function with the scan sets before (Figure

58 A) and after (Figure 58 B) manually removing the noise and artifacts from the teeth. Despite

manually removing the problem areas, there was no appreciable difference between the two

models.

Figure 55: The mandible was acquired with three sets of 360 scans. The first was made with the mandible in

anatomical position (A). The second and third were made with the lateral surface of the ascending ramus placed flat

on the part-gripper base, one set for the right side (B) and another for the left (C).

Figure 56: Much of the surfaces of the mandible were captured with the 360 scan sets. The teeth however had

numerous artifacts (a) and noise (b).

Figure 57: Artifacts in the scan data (A) were manually removed from individual scans by using the polyline tool

(B) to select problem areas (C) and remove them (D).

Figure 58: A water-tight model was created before (A) and after the manual removal of the problem areas in the

teeth (B). Despite the manual removal of artifacts and noise from the scan data, there was no appreciable difference

between the two models.

CHAPTER FIVE: RESULTS – APPLICATIONS

Once the scan data has been acquired, processed and exported as a 3D file, the resulting

data can be utilized to achieve a variety of virtual techniques that can be employed by the

forensic anthropologist. The most basic application of the 3D file is as a form of documentation

which can be reviewed with a computer at any time. More advanced techniques include

recording virtual measurements, creating illustrations with rendered stills, and performing virtual

reconstructions of fragmented or missing bone. This chapter will outline the software and

procedures used for performing these techniques.

Documentation

The most basic operation that can be performed with the 3D model once it has been

created is to save the file, label it in its own folder and add this to a library of scan files as a form

of digital curation. The model files can then be secured with a password protected ZIP folder or

external storage device, and sent to colleagues for additional examination. However, any user

attempting to open the file would require a program for viewing 3D models. In order to

circumvent this and standardize the viewing of the 3D model, the file can be embedded into a

PDF using Adobe® Acrobat® X Pro. Although a distinction was made between documentation

and visualization in Chapter 2, the creation of a 3D PDF serves to integrate both of these

applications.

3D PDF

A 3D PDF is an ordinary PDF file that contains an embedded 3D file which can be

opened with Adobe® Acrobat® Reader. However, viewing the 3D PDF may be limited by the

computer’s hardware based on the size and resolution of the model embedded within the PDF. A

graphics card with 3D capabilities is also required, and greater quantities of random-access

memory (RAM) allow the viewing of larger models without inhibiting computer performance.

Together, Adobe® Photoshop® CS5 and Adobe® Acrobat® X Pro are utilized in the production

of 3D PDFs. The creation of a 3D PDF is essentially a file conversion process, achieved through

changing the OBJ file to a U3D file and then saving the U3D file into a PDF (Figure 59).

Figure 59: A basic workflow illustrating the file conversion process for creating a 3D PDF from the scan data file

that was saved in the OBJ format.

O U3D in Adobe® A ® X Pr

Open U3D file by

changing file type to All

Files

Modify advanced

settings if needed

Save as PDF

O OBJ file in Adobe® Phot ® CS5

Right-click on 3D layer in Layers

pallette and export 3D layer

Save as U3D file type

Figure 60: In Adobe Photoshop CS5, the Layers palette is usually docked on the right side of the screen. To export

the OBJ as a U3D file, right-click on the 3D layer and click on Export 3D layer (highlighted in red). This brings up a

Save As screen, and allows the user to change the saved file format to a U3D file.

First, the OBJ file of the scan data is imported into Adobe® Photoshop® CS5. After this,

it can be saved as a U3D file by going to the Layers palette, right clicking on the 3D layer and

choosing ‘Export 3D layer’ (Figure 60).Once the U3D file has been saved, it is then opened in

Adobe® Acrobat® X Pro to be saved as a PDF. Before saving the 3D file in PDF format,

Adobe® Acrobat® X Pro allows the user to make modifications to the document using the

advanced settings checkbox, which brings up a menu with three tabs. Several features that are

useful to change in this menu are located in the Default Display Settings under the 3D tab (Figure

61).

Figure 61: The Advanced Options window allows many features to be modified before creating the PDF.

An important feature to change before creating the 3D PDF is the default lighting scheme.

The default lighting scheme is Lights from File, which will create a fixed highlight on the model.

The Headlamp light scheme is more preferable, causing whatever surfaces of the model are

facing the user to be illuminated interactively. In this menu it is possible to change the default

Lighting Scheme from ‘Lights from File’ to ‘Headlamp’, which will consistently light the

surfaces that are facing the viewer while interacting with the 3D object. Although this setting can

still be changed after the file has been created, it is efficient to create this as a default setting so

that the user does not have to change it every time the file is opened. Modifying the default

settings is especially useful if the file will be sent to lawyers, colleagues and other experts so that

the default shading is of a higher quality when the document is opened. Whether or not any

advanced settings have been chosen, the 3D PDF is created simply by saving the opened U3D

file as a PDF (Figure 62).

Figure 62: The scan data of the femur complete with texture, imbedded and viewed within a PDF file.

Issues

The final model of the femur that was produced in Chapter 4 (Fig. X) was used to

create a 3D PDF using the workflow outlined above. When opening the PDF created with this

file and attempting to rotate the model within the viewport, the rotation was delayed and the

computer had to load each time an attempt was made to move the model. After troubleshooting,

it became apparent that the high-resolution of the 3D model was causing the performance issue.

To correct this problem, the Simplify function in ScanStudio HD Pro™ was used to decrease the

amount triangles contained within the model. This feature is found by going to the Polish button

on the main toolbar and choosing Simplify (Figure 63). A specific area or the entire model can be

selected. In this case, the entire model was selected for simplification by clicking ALL (Figure

64). Note that the model must also be fused prior to simplification. The process of simplifying a

model in ScanStudio HD Pro™ requires experimentation since the software only allows the user

to specify a general tolerance value for the simplification and not a specific target triangle count.

It is also possible to simplify the model too much, which results in loss of accuracy. To efficiently

organize the simplification process for determining an optimal model resolution, it is important

to document the original quantity of triangles contained within the model and compare this with

the triangle counts that result from simplification using different tolerance values. The total

number of triangles can be viewed by going to File in the main toolbar and clicking on Model

Information (Figure 65), or by using the keyboard shortcut Ctrl+Shift+I.

Figure 63: The Simplify function is found under the Polish tab in the main toolbar.

Figure 64: The simplification toolbar within ScanStudio HD Pro™. The entire model is selected by clicking the

ALL button. The simplification level can then be modified by moving the Simplify slider or by typing in a number

to change the tolerance value.

Figure 65: The total number of triangles for the entire model or the attached data is identified in the Model

Information window, which can be accessed under File in the main toolbar or simply by using the keyboard shortcut

Ctrl+Shift+I.

If the model has been fused, then it should be displayed in the green section of the model

list located at the bottom of the ScanStudio HD Pro™ interface. This area is known as the

Attached data section, and the corresponding triangle count is listed at the bottom of the Model

Information window. After the triangle count has been noted, several simplification operations

should be performed using different tolerance values. The tolerance value and resulting triangle

count should be documented and performed until there are visible distortions, or the model loses

a significant amount of detail. The tolerance value that was used prior to the loss in surface detail

should be an appropriate level of simplification for the creation of a 3D PDF. To demonstrate the

loss of resolution, the original fused model of the femur was simplified with five different

tolerance settings: 0.0125”, 0.0250”, 0.0500”, 0.1000 and 0.2000”. A rendered still image was

then created in ZBrush™ 4r4 and Adobe® Photoshop® CS5 for comparison. Figure 66 shows a

shaded view of the original femur and additional copies which were processed with different

tolerance values. Although the features remain largely intact, faceting on the head of the femur is

visible with a tolerance value of 0.1000” and is especially pronounced with a value of .2000”.

This faceted appearance is the result of larger triangles that were produced during simplification

with higher tolerance values. Figure 67 shows a wireframe view, which illustrates the triangle

density of each model listed in Table 1.

Figure 66: This image was created with ZBrush™ 4r4 and Adobe® Photoshop® CS5 to illustrate the loss of detail

that results from greater tolerance values. The original model without simplification is located at the top, with

increasing tolerance values resulting in higher simplification. Note the faceting of the head of the femur with greater

tolerance values (0.2000”).

Figure 67: The simplification process is more apparent with a wireframe view of the femur, illustrating the density

of triangles within each mesh.

Table 1: A chart illustrating the amount of triangles contained in the original, and for each tolerance value that was

processed with the Simplify feature in ScanStudio HD Pro™.

Original

1,140,714 triangles

0.0125”

270,658 triangles

0.0250”

118,847 triangles

0.0500”

52,942 triangles

0.1000”

21,833 triangles

0.2000”

9,384 triangles

Choosing the appropriate tolerance value depends on the intended purpose of the PDF

file. If the file will be used simply for presentations and education, the higher simplification

levels should be acceptable. If the PDF will be sent to a colleague for measurement or for

presentation in court, then lower simplification levels should be appropriate if the original

resolution of the model inhibits computer performance. Regardless, the original resolution model

should always be maintained for documentation purposes.

Measurement

Although the NextEngine ScanStudio HD Pro™ software is able to measure surface area,

and volume with additional software, it does not have a function for collecting basic point-

topoint measurements. This deficiency is easily overcome with the use of the 3D measurement

tool accessible within a 3D PDF document. The 3D measurement tool is found in the drop down

menu that is accessed by clicking on the arrow next to the left-most icon in the 3D viewport

toolbar (Figure 68).

Figure 68: The 3D measurement tool is accessed from the drop down menu on the left side of the toolbar.

Once the 3D measurement tool has been selected, a measurement is recorded by clicking

on the model to choose the first point and then clicking again to select a second point. The user

can navigate within the viewport before selecting the second point if the required location is not

visible in the current view. This is done by holding one of several keys while clicking and

dragging outside of the model, but while still remaining within the 3D window. The Alt key is

used to rotate the model, the Shift key is used to pan the model around the screen, and the

combination of Alt+Shift is utilized to zoom in and out. After the second point is selected, the

resulting measurement is shown within the viewport and can be oriented away from the model by

moving the cursor and clicking once (Figure 69).

Figure 69: An arbitrary measurement being recorded from the femur with the 3D measurement tool within a PDF

opened in Adobe® Acrobat® Reader.

After an individual measurement is recorded, it can be stored and cleared by clicking on

the default view icon in the main tool bar. The measurements taken are saved as Views, which

can be accessed from the corresponding drop-down menu in the main toolbar. These

measurement views can then be relabeled by clicking on Manage Views at the bottom of the

drop-down menu. In order to retain these measurement views within the document, the PDF must

be re-saved.

Issues

The virtual measurement of the human cranium has undergone extensive scrutiny in

testing the reliability of this method when applied to CT data (Hildebolt et al., 1990; Richtsmeier

et al., 1995; Weber et al., 1998), but there have only been limited studies assessing the reliability

of this approach when applied to laser scan data (Park et al., 2006; Scholts et al., 2010). As

mentioned in Chapter 2, Park et al., (2006) and Scholts et al., (2010) did not use the standard set

of measurements used by forensic anthropologists when measuring the human cranium that are

provided by Buikstra & Ubelaker (1994). Thus, the reliability of recording virtual measurements

from laser scan data of the human cranium has not been established for use in forensic

anthropology. In order to address this situation, the standard craniometrics used for FORDISK

(Ousley & Jantz 2005) classification were used in this study, which are outlined by Buikstra &

Ubelaker (1994). These measurements were then collected with manual and digital techniques

for comparison. The manual measurements were recorded from the cranium using both spreading

and sliding calipers, and the digital measurements were acquired using the 3D measurement tools

within the PDF document, which was embedded with a 3D model created from scan data of the

same cranium. Before creating the PDF, the cranium model was simplified with a tolerance of

0.0250”. This lowered the triangle count from 3,324,694 triangles to a more manageable 493,356

triangles. The texture information was also discarded by isolating the OBJ file into its own folder

prior to converting it to a U3D file within Adobe® Photoshop® CS5.

After the measurements were collected using the 3D measurement tool within the PDF,

they were organized into a chart for comparison (Table 2). Note that the maximum alveolar

breadth was not collected because the palette was resorbed. Upon observation, several

measurements were found to be exactly the same and the measurements that were different

deviated by only one or two millimeters. To further investigate the reliability of the digital

technique, each set of measurements was analyzed through FORDISC 3.0 (Ousley & Jantz

2005), using all groups selected, and then only with male groups selected. The results of applying

multigroup discriminant function analysis to both sets of measurements indicated a classification

of the cranium as a Chinese male. When assessed based on sex and ancestry the manual

measurements resulted in a posterior probability of 0.371 and a typicality of 0.593. The digital

measurements resulted in a posterior probability of 0.652 and a typicality of 0.615. When

assessed based only on sex, the manual measurements resulted in a posterior probability of 0.588

and a typicality of .640, and the digital measurements resulted in a moderately high posterior

probability of 0.706 and typicality of 0.638. The analysis of both manual and digital

measurements resulted in the same classification, suggesting the reliability of the digital

techniques. Of special note is that the digital measurements resulted in a slightly higher

probability for the resulting classification.

Table 2: The results of the FORDISC measurements taken with manual and digital techniques.

FORDISC

Measurements

Manual

Digital

Difference

Maximum length (g-op)

174 mm

176 mm

2 mm

Maximum breadth (eueu)

139 mm

137 mm

2 mm

Bizygomatic breadth (zy-

zy)

130 mm

129 mm

1 mm

Basion-bregma (ba-b)

142 mm

143 mm

1 mm

Cranial base length (ban)

95 mm

96 mm

1 mm

Basion-Prosthion length

(ba-pr)

96 mm

95 mm

1 mm

Max. Alveolar breadth

(ecm-ecm)

Max. Alveolar length (pr-

alv)

50 mm

n/a

Biauricular breadth

(auau)

116 mm

117 mm

1 mm

Upper facial height

77 mm

n/a

Minimum frontal breadth

(ft-ft)

88 mm

89 mm

1 mm

Upper facial breadth

(fmt-fmt)

98mm

98 mm

n/a

Nasal height (n-ns)

55 mm

n/a

Nasal breadth (al-al)

25 mm

n/a

Orbital breadth (d-ec)

38 mm (left) 38 mm

(right)

38 mm (left) 39 mm

(right)

n/a & 1 mm

Orbital height

37 mm (left) 39 mm

(right)

39 mm (left) 40 mm

(right)

2 mm & 1 mm

Biorbital breadth (ec-ec)

94 mm

95 mm

1 mm

Interorbital breadth (d-d)

20 mm

n/a

Frontal chord (n-b)

117 mm

n/a

Parietal chord (b-l)

113 mm

n/a

Occipital chord (l-o)

98 mm

96 mm

2 mm

Foramen magnum length

(ba-o)

34 mm

36 mm

2 mm

Foramen magnum

breadth

29 mm

n/a

Mastoid length

28 mm (left) 28 mm

(right)

29 mm (left) 29 mm

(right)

1 mm & 1 mm

Illustration

Rendered Still Images

The 3D model processed from laser scan data can be rendered to create 2D illustrations

for use in reports, presentations, and publications. These rendered still images are the result of the

rendering process, which is a digital method for creating high-quality or photorealistic 2D images

from 3D models through the simulation of lighting, shading, material and texture information.

Basic rendered images can easily be made by importing a file into ZBrush™ 4r4 and using the

Best Preview Render (BPR) feature (Figure 70).This is done by clicking Import in the Tool

palette and choosing an OBJ file. The Best Preview Render feature can then be accessed by

going to the Render palette, and clicking on the BPR icon under the BPR RenderPass subpalette

(Figure 71).

Figure 70: A still image rendered with shadows using the Best Preview Render feature inside of ZBrush™ 4r4.

Figure 71: The Best Preview Render feature within ZBrush 4r4 can be accessed by clicking the BPR button in the

Render palette, highlighted here by the red square.

A quicker alternative for using the BPR feature is to use the key board shortcut Shift+R.

The resulting image can then be saved either as a screenshot by pressing the Print Screen key or

by going to the Document palette and using Export to save the image as a PSD or other image

format. The shadows in the image can be controlled by going to the Light palette and

manipulating the colored dot located on the sphere object (Figure 72). This signifies the location

of the light source, and the changes will become apparent after refreshing the render. Additional

lights can be added by clicking one of the light bulb icons next to the sphere object within the

light palette.

Figure 72: Within ZBrush 4r4, the shadows can be manipulated by moving the light source. This is done by moving

the colored dot located on the sphere object within the Light palette, outlined here in red.

The texture information contained within the MTL file that is exported with OBJ files in

the ScanStudio software cannot be imported into ZBrush™ 4r4. Instead, a new material can be

applied to the entire model by clicking on the Material icon in the left-hand palette and making a

selection. The user can also create their own materials, or download materials created by the

ZBrush™ community from the Pixologic MatCap Library

(http://www.pixologic.com/zbrush/downloadcenter/library/). The MatCap Skeleton material was

used with the model in Fig. X. Because the materials in ZBrush™ 4r4 do not represent the actual

color of the object, the rendered stills created using this technique should only be used in

illustrations where color information is not necessary. If texture information in ZBrush™ 4r4

renders is required, it is possible to manually texture the model with photographic references

using the SpotLight feature, but this technique will not be implemented here.

Virtual Reconstruction

When used in conjunction with laser scanned images, the digital sculpting capabilities of

ZBrush™ 4r4 offer a non-invasive alternative to traditional reconstruction techniques for

assembling and recreating damaged or fragmented bone. The virtual reconstruction technique

that is implemented in this project is a manual technique and does not have the scientific

accuracy of other studies that utilize reverse engineering methods such as the one performed by

Benazzi et al. (2009). The example used here to illustrate the digital reconstruction technique in

ZBrush™ 4r4 was a resin cast of an intact cranium and two mandible fragments (Figure 73) from

an actual forensic case purchased from Bone Clones Osteological Reproductions. The cast was

used in lieu of an actual human skull because a fragmented example was unavailable for this

study. The information that arrived with the replica states that the original cranium and mandible

fragments belonged to a female of African ancestry who suffered from two shotgun wounds to

the occipital bone. The mandible was fragmented into at least three pieces, of which only two

were present. The left zygomatic arch was also missing.

Figure 73: A resin cast of a cranium and two mandible fragments were scanned to use as an example for the manual

digital assembly and reconstruction.

The cranium and mandible fragments were scanned and processed using the protocols

outlined in Chapter 4, which resulted in three files that were exported in the OBJ file format.

First, the cranium file was imported into ZBrush™ 4r4 by going to the Tool menu and clicking

import. In ZBrush™ 4r4, a model file is described as a Tool, and the individual selectable

components that comprise the model are called Subtools. After the cranium was opened in the

program, it was reoriented using the offset and rotation sliders in the Deformation palette (Figure

74).

Figure 74: Subtools can be moved and rotated with precision by using the Offset and Rotate sliders in the

deformation palette.

In order to assist in the orientation of the cranium, a series of planes were added as

subtools by clicking on Append in the subtool palette and choosing the Plane3D tool (Figure

75A&B). In order to render both sides of the plane visible when rotating the tool, it is necessary

to select the plane subtool and click on Double in the Display Properties sub-palette within the

main tool palette. The first Plane3D was placed in the frontal plane by default (Figure 75 A). The

second plane was created by using the keyboard shortcut Ctrl+Shift+D to duplicate the first

plane. This new plane was then rotated 90° in the Y axis using the Rotate slider in the

Deformation palette so that it was oriented in the sagittal plane (Figure 75 B). These planes

served as a reference for orienting the cranium in a symmetrical manner, with the sagittal suture

and craniometric points rhinion and nasion lined up with the Plane3D oriented in the sagittal

plane, and the features at the base of the skull aligned with the frontal plane.

Figure 75: The appended Plane3D was oriented in the frontal plane by default. The features at the base of the

cranium were then used along with the reference plane to orient the cranium into anatomical position (A). A

duplicate of the first Plane3D was rotated 90° to orient it in the sagittal plane. The craniometic points nasion (a) and

rhinion (b) were then lined up with the reference plane (B).

In order to add the mandible fragments to the cranium, it is necessary to append another

subtool, which will be replaced by the imported file. It does not matter what subtool is appended,

as it is simply a placeholder for the imported file. In this instance, two subtools were appended,

one for each of the mandible fragments. Once the mandible fragments were added as subtools,

they were moved and rotated into the correct position. This proved to be a challenge using the

offset and rotate sliders in the deformation palette. Instead, the Transpose feature was used to

move and rotate the mandible fragments. The Transpose feature is accessed by clicking on the

Move, Scale or Rotate buttons in the main toolbar (or by using the Hotkeys ‘W’ ‘E’ and ‘R’,

respectively), clicking on the subtool and dragging the action line off of the model. The action

line can be conformed to a straight-line by also holding shift. The model is then moved or rotated

by clicking and dragging on one of the three rings found on the action line. The scale feature was

not used so that the actual proportions between the cranium and mandible fragments would be

maintained. After considerable effort, the mandible fragments were visibly aligned with the

cranium and placed into anatomical position (Figure 76).

Figure 76: The cranium and mandible fragments oriented in anatomical position.

Once the cranium and mandible fragments were oriented in anatomical position, the

reconstruction of the missing areas was performed using the free-form modeling tools inside of

ZBrush™ 4r4. There are many tools inside of ZBrush™ 4r4, and therefore different ways that the

reconstruction can be performed. The method outlined here is simply the one developed and

employed by the author based on personal experience and preference with the program. Each of

the reconstructed areas was created using the Dynamesh feature inside of ZBrush™ 4r4.

Dynamesh is a unique solution for free-form modeling complex geometry without having to

worry about topological problems such as polygon stretching.

The reconstruction of the left zygomatic arch began with a Sphere3D. A Sphere3D is one

of many tools inside of ZBrush™ 4r4 known as primitives, which can be manipulated using the

Masking and Initialize functions to create complex geometric objects. However, unmodified

primitives are not very useful for creating organic shapes such as bone. Instead, the Sphere3D is

converted to a PolyMesh3D by clicking the Make PolyMesh3D icon at the top of the Tool

palette. Doing this will enable sculpting, and should be performed before appending the object to

the cranium subtool. Once converted to a PolyMesh3D, the Dynamesh feature is then activated

by clicking its respective icon in the Geometry subpalette within the main Tool palette. This will

restructure the geometry of the object to make it more uniform, and can be re-applied to the tool

as it is being modeled by pressing CTRL and then clicking and dragging within the viewport

outside of the model. The Move brush was used to roughly shape the sphere into the form of the

missing fragment. The standard Smooth brush was used in conjunction with the Move brush to

help smooth out areas during the sculpting process. It is accessed by holding Shift while using

any of the sculpting brushes. These two brushes were used until the replacement fragment looked

believable, while visually referencing the contralateral zygomatic arch which was still intact

(Figure 77).

Figure 77: The zygomatic arch was reconstructed using Dynamesh, and the Move & Smooth sculpting brushes.

The holes in the occipital bone were reconstructed with a Sphere3D tool with Dynamesh

active. The spheres were then placed within the cranium and moved until the surfaces roughly

matched the borders of the cranium. The Inflate brush was then used to form the surface of the

sphere that was visible through the hole until it matched the surfaces of the skull without

appearing over-inflated or caved-in (Figure 78).

Figure 78: The missing areas that resulted from shotgun wounds to the occipital bone were filled in using Dynamesh

and the Inflate brush.

The teeth and mandible were the most difficult areas to reconstruct. Like the other

reconstructed areas, the mandible fragment was created from a Sphere3D with Dynamesh

activated, except that the symmetry feature was used. Symmetry is activated by pressing X on the

keyboard, and the axis that symmetry will be mirrored on can be changed in the Transform

palette. The default axis of symmetry in ZBrush™ 4r4 is in the X axis, and in this instance

matches the bilateral symmetry of the cranium and mandible. Using symmetry, the Dynamesh

sphere was modeled to fit the missing area while visually referencing intact mandibles during the

process (Figure 79). The teeth were modeled previously as an anatomy study, and were imported

and placed on the cranium and mandible (Figure 80 B).

Figure 79: The mandible was reconstructed using Dynamesh with symmetry activated in the X-axis.

Once all of the missing fragments have been remodeled, they can be fused with the scan

data in order to unite the different subtools into a single part using the Remesh function. This is

especially useful if the model will be 3D printed. First, the reconstructed areas of the cranium

were fused to the bone by first toggling the visibility so that only these fragments and the

cranium were visible because the ReMesh function is applied to all visible subtools. Visibility is

toggled by clicking on the eyeball icons displayed to the right of the respective tools in the

subtool palette. The Remesh is then performed by clicking ReMesh All in the subtool palette.

This function includes a symmetry feature located on the ReMesh All icon, but it should be

turned off in this instance. The resolution of the ReMesh can be modified by changing the Res

slider. Once a new subtool has been created using the ReMesh feature, the ProjectAll icon below

ReMeshAll is used to project the surface details of the previous subtools onto the new subtool.

This should be used in conjunction with the Res slider until the ReMesh function creates a high

enough resolution subtool that can capture the level of detail of the existing subtools with

ProjectAll (Figure 80 C)

Figure 80: The scan data of the fragmented cranium and mandible were imported and assembled in ZBrush™ 4r4

(A). Missing areas were then reconstructed by digitally sculpting them in (B). Once the reconstructed areas were

created, the model was remeshed (C).

CHAPTER SIX: DISCUSSION & CONCLUSION

Discussion

3D digitization has become an integral part of many disciplines within the field of

anthropology, from the use of CT scanners in paleoanthropological studies (Aiello et al., 1998;

Hardwood-Nash 1979; Seidler et al., 1992), to the application of long-range laser scanning

technology in the documentation of excavation sites in archaeology (Lambers et al., 2007). As

the advantages of 3D digitization technologies continue to be developed, more and more

researchers from a variety of subfields are becoming interested in implementing this technology

into their practice. Forensic anthropology is one such area where 3D digitization is beginning to

be implemented, albeit slowly. In consideration of this pace, this research project was pursued in

hopes of increasing the momentum at which these technologies will be adopted within this field

through an investigation of the potential applications. To investigate the practicality of

implementing 3D digitization technologies within a forensic anthropology lab, an affordable laser

scanner was chosen to conduct this study. The study was divided into two primary components.

First, an exploratory documentation of protocols for scanning different human bone types and

techniques for processing the data was constructed in an effort to standardize the process of

scanning human bone. The second section was an exploration of practical applications for using

the scan data within the field of forensic anthropology. Many challenges were encountered

during this investigation, but the advantages of using the scan data appear to be worth the effort.

The challenges in scanning the different bone types and the potential uses for the scan data are

outlined and reviewed in this chapter.

Scanning Protocols

The challenges in scanning different bone types such as the vertebra and scapula, as well

as the potential uses for the scan data are outlined and reviewed in this chapter. The issue

presented in scanning the L5 vertebra was that the borders of the vertebral foramen were

unobtainable with the laser scanner. When attempting to process the scan data that was obtained,

the software completely filled in this area when attempting to create a water-tight model with the

fuse function (Fig. X). This problem could be resolved by using a bridge function to break up the

missing data into smaller holes which could be filled individually while maintaining surface

curvatures, but this capability is not available within the ScanStudio HD Pro™ software.

The problem in laser scanning the scapula involved the thin borders of the subscapular

and infraspinous fossae. The infrared lasers of the scanner passed through the translucent

surfaces of the fossae, which appeared as the inversion of the surface normals within the Scan

Studio HD Pro™ software (Fig. X). The application of talc powder is suggested in the

NextEngine Support manual for scanning translucent and reflective surfaces. Talc power was

applied to the surfaces of the fossae using one of the powder brushes supplied with the scanner,

but this did not result in resolving the problem. Additionally, a scapula that had been painted for

teaching purposes was also scanned, but the paint also did not prevent the problem.

Both the clavicle and sternum were simple to scan and did not present any particularly

unique challenges. The ease in scanning these bones lies in their small size and relative simplicity

of form.

Despite the unique morphology of the os coxae, scanning the bone was manageable. The

only issue arose with certain areas that appeared to be damaged, where the cancellous bone was

exposed. These areas appeared as holes in the scan data, but were filled without too much

distortion of the features when fused as a water-tight model. Since damaged areas can be a sign

of trauma, these areas are particularly important during an investigation and this issue should be

considered by the forensic anthropologist. This validates the findings of Slizewski & Semal

(2009), who also mention that the distortion issue with cancellous bone also arises with CT data.

Scanning the rib was a challenge because it was difficult to secure it in place with the

part-gripper. This illustrates an issue with the scanner hardware, namely the design of the

partgripper. The part-gripper was modified by adding an additional supporting prong from

another part-gripper to increase the range of support. Despite this, it still took considerable effort

to stabilize the rib for scanning. A design feature that would enhance the ability to scan objects

that are not self-supporting could be the addition of rubberized clamps that would secure the

bone in place without damaging it.

The femur’s unique characteristic of being the longest bone in the human body is what

made it complicated to scan. Although the ScanStudio HD Pro™ software has a Wide and

Extended scan setting for digitizing larger objects, these settings result in lower resolution

models with less detailed anatomical features. Because of this, the Macro setting was maintained

and the scanner itself was raised to varying heights for capturing the entire length of the femur. It

is possible to mount the scanner on a tripod, but in this case it was simply placed on a stack of

large textbooks.

The cranium was the most difficult bone in the sample to collect. It required extensive

repositioning and scans to acquire as many surfaces as possible, yet there were still multiple

areas which could not be placed within the line of sight of the scanner, and thus could not be

acquired. These included the surfaces within the nasal aperture which include the vomer and

nasal conchae, the internal surfaces of the cranial vault, the inner medial surfaces of the

zygomatic arches, and many of the small intricate surfaces at the base of the skull. When creating

a water-tight model, the natural voids within the skull become closed. These areas include the

foramen magnum, orbital fissures and all canals, fossae and foramen. Despite the distortion of

these areas, the standard craniomentric measurements used by forensic anthropologists were

successfully applied to the digital model of the cranium. If the distortion of these areas is

considered to be too significant to ignore, the scan data can be fused without hole-filling to retain

the natural openings in the skull. However, the model with holes in the mesh will not be able to

be 3D-printed. If a 3D model of the human skull with every anatomical feature intact is required,

then CT scanning would be necessary.

Scanning the bone of the mandible did not present any particular difficulties. On the other

hand, the teeth were problematic to scan on both the mandible and cranium. This reaffirms the

findings of Slizewski & Semal (2009). The difficulty in scanning the teeth arises from the minute

intricacy of the occlusal surfaces, in addition to the interpoximal surfaces which were not within

the scanner’s line of sight. The inability to accurately acquire these surfaces is compounded by

the slight translucency of the teeth, which absorbs the laser beam (Slizewski & Semal 2009) and

results in surface noise. Even if the noise is manually removed from the scans, the missing data

that results from the inability to scan the interproximal surfaces between the teeth causes the

software to fill these areas as holes, resulting in significant distortions. Although Slizewski &

Semal (2009) point this out, there are no solutions provided other than the use of a CT scanner.

Scan Data Processing

Processing the scan data once it has been acquired is a relatively straightforward task

using the ScanStudio HD Pro™ software. Conversely, the stream-lined and user friendly

software lacks many advanced features for manual processing of the scan data. These advanced

features can aid in overcoming some of the issues that were encountered during the scanning

process. One such feature is the ability to bridge polygons to break up large holes into smaller,

more manageable holes. This would be useful in the instance of the vertebra, where the missing

data within the vertebral foramen created a large hole which was filled over by the software when

making a water-tight model using the fuse function. The bridge tool could be used in this

instance to break up the large hole into smaller holes which could be filled individually while

maintaining the surface curvature within this area.

Bridge tools are commonly available in 3D modeling software such as Autodesk® 3DS

Max® and Maya®, but these programs usually cannot handle large scan data sets. Geomagic

Studio® is a scan data processing software that contains a bridge feature, in addition to many

other features not found within ScanStudio HD Pro™. The cost of licensing this software was

prohibitive for utilizing it in this project, but it is strongly recommended for integration into a 3D

digitization workflow. If the cost of licensing the software becomes more affordable, it may be a

viable asset for processing scan data of human bone in forensic anthropology. Alternatively,

researchers and investigators interested in advanced scan data processing could seek out other

departments and institutions that have licensed this software, in collaborative efforts.

Scan Data Applications

Once the challenges of obtaining and processing the scan data have been overcome, there

are many useful applications in which the scan data can be put to use.

Documentation

Using the scan data as a form of digital documentation allows the forensic anthropologist

to review the remains at any time (Park et al., 2006). This is advantageous if the remains become

damaged or are no longer available for physical examination (Tocheri, 2009). In addition, the

digital replica of the remains can be sent to colleagues and other experts for analysis almost

instantly over the internet, without having to pay for the costs of transportation (Davy-Jow et al.,

2012). The viewing of the scan data can be standardized and made more convenient by

embedding the file into a PDF document using Adobe® Photoshop® CS5 and Acrobat® X Pro.

Measurement

In conjunction with the creation of a PDF file embedded with the scan data is the ability

to collect virtual measurement from the digital model. The reliability of collecting digital

measurements was assessed through a basic study performed in Chapter 5. Current publications

assessing the reliability of recording virtual measurements from laser scans of the human

cranium have implemented robust statistical methods (Park et al., 2006; Sholts et al., 2010), but

have not utilized the standard landmarks used in craniometry by forensic anthropologists. The

study performed in this project employed the standard landmarks and processed the results with

FORDISC 3.0 (Ousley & Jantz, 2005) using multigroup discriminant function analysis, but

unlike Park et al., (2006) and Sholts et al., (2010) it did not assess the reliability based on inter-

and intra-observer error. The sample size was also limited to a single cranium. Future studies

investigating the reliability of virtual measurements collected from laser scans of human crania

should use the standard measurements, in addition to assessing inter- and intra-observer error

with a larger sample size.

Illustration

Utilizing rendered stills can be a useful supplement to photography for creating visual

documentation of human bone. In conjunction with the ability to rotate the model to view any

surface of the bone, the rendered stills enable the user to highlight specific areas for illustration.

These images can then be used in lectures, research publications and court presentations in lieu

of exhibiting actual images which could be potentially disturbing to some audience members,

such as the victim’s family (Komar et al., 2012).

Virtual Reconstruction

Assembling and reconstructing fragmented bone through the use of a laser scanner and

3D modeling software allows the remains to be maintained in their original state (Fantini et al.,

2008). This is advantageous over traditional methods of assembly and reconstruction that use

adhesives, which can be potentially destructive and irreversible. Once the fragmented bones have

been assembled and reconstructed, the digital model can be made physical through 3D printing

(Zollikofer & Ponce de Leon, 2005).

Other Issues to Consider

In summary of the results for both scanning, processing and exploring the uses of the

scan data, a laser scanner was determined to be an important addition to a forensic anthropology

lab. Although some of the issues encountered in establishing protocols for scanning different

bone types could not be resolved, the majority of the bone types that were scanned resulted in

reliable representations of the original objects.

Some of the limitations that were encountered are inherent with the technology. Infrared

lasers will penetrate non-opaque surfaces such as teeth and the thin surfaces of the scapular

fossae. A potential solution for this would be the ability to alter the laser intensity of the scanner.

Lowering the intensity may decrease the penetration of the lasers through the thin surfaces and

allow them to be collected. The ScanStudio HD Pro™ software does not allow for user input of

laser intensity values, but there are programs that contain this feature such as Polygon Editing

Tool (PET) which is used in conjunction with the Minolta VIVID 9i non-contact 3D Digitizer.

The Minolta VIVID 9i is nearly five times the price of the NextEngine™ and may not be a

practical alternative for overcoming the issue of laser intensity when scanning non-opaque bone

surfaces. If NextEngine™ upgraded their software for more advanced features such as the ability

to input specific values for modifying laser intensity, this issue could be resolved. However, it is

possible that in the near future, other affordable laser scanners with more technical proprietary

software will surface on the market.

An additional restriction of laser scanners is the line-of-sight, which allows only the

surfaces within the visible range of the scanner to be collected. This is in contrast to CT

scanning, which will collect even the internal surfaces of an object. Due to the restricted line of

sight of the laser scanner, the internal borders of the vertebral foramen and many surface within

the cranium such those within the nasal cavity could not be acquired with the use of a laser

scanner.

Considering that many of these surfaces are not necessarily important to the forensic

anthropologist, the issue may be an acceptable compromise considering the ability to digitize

skeletal material in an affordable way. There are a few ways that the line-of-sight of the scanner

could be increased, but would be dependent on additional hardware. A new part-gripper design

that allowed more dynamic manipulation of the object being scanned, while still maintaining the

objects stability could allow for more surfaces to be captured than is allowed by the fairly limited

setup of the NextEngine™ scanner. Another method to increase line-of-sight would be to mount

the scanner on a tri-pod that allows for rotation in multiple axes.

Despite these limitations, the complete models that were obtained can be used in a variety

of useful ways. The ability to permanently curate case materials in digital form by embedding the

3D file into a PDF enables the forensic anthropologist to review them at any time and allows

them to be sent to colleagues and other experts for evaluation without involving costs and

procedures for transportation. Additionally, there are 3D measurement capabilities afforded by

the PDF. When this tool was employed to collect the standard cranial measurements used by

forensic anthropologists and compared with analogous measurements taken with calipers, the

results illustrated the reliability of using digital measurements taken from the scan data. The scan

data can also be used to create 2D illustrations analogous to photographic images, with the

Students also viewed