Order 1106099: migraine

S. Supek and A. Sušac (Eds.): Advances in Biomagnetism – BIOMAG2010, IFMBE Proceedings 28, pp. 183–186, 2010. www.springerlink.com

Quantification of the Time and Frequency Signatures of Visual Cortical Activation in the Developing Brain: A Study with MEG and Wave-Cross Spectrogram

Xinyao Guo, Jing Xiang, Yangmei Chen, Lu Meng, Xiaoshan Wang, and Yingying Wang

MEG Center, Department of Neurology, Cincinnati Children’s Hospital Medical Center, Cincinnati, USA

Abstract— Our previous studies have demonstrated that Morlet-wavelet transform with an extra large sigma value could precisely determine the frequency signatures of neuro- magnetic signals. Unfortunately, the increase of frequency sensitivity is associated with a decrease of temporal resolution. To solve this problem, we have developed wave-cross analysis to quantify neuromagnetic signals with both high temporal and frequency resolutions. The objective of the present study is to measure the time and frequency signatures of visual cortical activation in children using this new method. Twelve healthy children were studied with a whole-head magnetoencephalo- graphy (MEG) system. Visual evoked magnetic fields (VEFs) were evoked with full-field pattern-reversal checks. MEG data were transformed from time-domain to frequency domain using wave-cross spectrogram. Neuromagnetic sources were volumetrically localized with a wavelet-based beamformer. Three response peaks were identified at 73±7 ms (M75), 111±8 ms (M100) and 149±12 ms (M145). The latency of M75 and M100 decreased with age. The amplitude of M75 decreased with age while the amplitude of M100 increased with age. The amplitude ratio of M100/M75 increased significantly with age. The frequency bands up to 300 Hz have been identified. The frequency in the M73 appeared higher than M145, the later responses. In addition, the frequency signatures of the neuro- magnetic signals also changed with age. The temporal and frequency signatures of the development of visual function in childhood are noninvasively quantifiable with MEG and Wave-cross. Our preliminary data have confirmed that wave- cross, a new time-frequency analysis method, could precisely determine the frequency and temporal signatures of brain activation. The results lay a foundation for quantitative identi- fication of developmental delay and/or abnormalities of visual function in children with brain disorders.

Keywords— Magnetoencephalography, visual cortex, chil- dren, Wave-cross spectrogram, visual evoked magnetic fields.

I. INTRODUCTION

Recent advances have found that noninvasive biomarkers are critical to the study of functional brain maturation and developmental delay in children. Magnetoencephalography (MEG) is a new technology for detecting neuromagnetic fields associated with electric brain activities. It is generally accepted that MEG has both a high spatial and temporal resolution. Visual evoked magnetic fields (VEFs) in the

visual cortex in response to a pattern reversal visual stimu- lus have revealed several detectable components. Since visual evoked responses correlate to myelination and synap- tic transmission, VEFs are potential biomarkers for evalua- tion of visual function. The latency and amplitude of VEFs are commonly used to analyze the VEFs. However, the frequency signatures of VEFs are rarely used for evaluating visual function. Since brain activation changes with time, the conventional Fourier transform may not suitable for the study of the frequency signatures. Recent years, wavelet transform has been used in the study of brain function. Our previous studies have demonstrated that Morlet-wavelet transform with an extra large sigma value could precisely determine the frequency signatures of neuromagnetic sig- nals. Unfortunately, the increase of frequency sensitivity is associated with a decrease of temporal resolution. To solve this problem, we have developed wave-cross analysis to quantify neuromagnetic signals with both high temporal and frequency resolutions. The objective of the present study is to measure the time, frequency as well as amplitude signa- tures of visual cortical activation in children using the new method.

II. SUBJECTS AND METHODS

A. Subjects

Twelve healthy normal children (age: 6-17 years, mean age: 11; 6 female and 6 male) were studied with MEG. These children were recruited from the surrounding Cincin- nati area. Since these children were developmentally normal and had no irregular findings reported from MRI, we con- sidered that they were representative of the normal popula- tion. A written informed consent, at Cincinnati Children’s Hospital Medical Center (CCHMC), was obtained from the parent/legal guardian of each child. Inclusion criteria for participation were: (1) healthy, without history of neuro- logical disorder, psychiatric disease, or brain injury; (2) normal hearing, vision, and hand movement; (3) current age between 6 years – 17 years old. Exclusion criteria for par- ticipation: (1) a child could not keep still; (2) a child could not look at a fixing point or did too much eye blink; (2)

184 X. Guo et al.

IFMBE Proceedings Vol. 28

subjects with learning and /or speech/language disability; (3) subjects with unidentifiable magnetic noise.

B. Methods

Stimuli: Visual stimuli were generated by BrainX (Xiang et al., 2001), a program based on DirectX (version 9; Mi- crosoft Inc., Redmond, WA, USA). The visual stimulus was a black and white pattern reversal checkerboard with a yel- low dot fixated in the center. The visual stimuli were pro- jected through a small hole in the magnetically shielded room (MSR) via an XGA portable multimedia projector (model PLC-XP41, Sanyo Electronics Ltd., Chatsworth, CA). Stimuli were projected onto a translucent screen lo- cated approximately 32 cm from the subject’s nasion. The visual angle of the checkerboard (the visual field diameter) was 30 degree; the visual angle of each square (the check) within the checkerboard was 1 degree and 15 min. The luminance of the white areas was about 80 cd·m−2. Rever- sal rate between the white and black checker was 1 Hz.

MEG Recording: The supine position, his or her arms resting on either side, during the entire procedure. The sam- pling rate of the MEG recording was 6000 Hz per channel. The MEG signals were recorded in a magnetically shielded room (MSR) using a whole head CTF 275-Channel MEG system (VSM MedTech Systems Inc., Coquitlam, BC, Can- ada) in the MEG Center at CCHMC. Before data acquisition commenced, three electromagnetic coils were attached to the nasion, left and right pre-auricular points of each sub- ject. These three coils were subsequently activated at differ- ent frequencies for measuring each subject’s head position relative to the MEG sensors in real-time. The data was re- corded with a noise cancellation of third order gradients and without on-line filtering. One hundred trials of MEG data were recorded for each subject. Subjects were asked to keep still. If the head movement during one recording was be- yond 5 mm, the dataset would be indicated as bad and an additional trial would be recorded. The entire MEG re- cording was monitored with an audio-video system.

MRI Scan: Three-dimensional Magnetization-Prepared Rapid Acquisition Gradient Echo (MP_RAGE) sequences were obtained for all subjects with a 3T scanner (Siemens Medical Solutions, Malvern, PA). Three fiduciary points were placed in identical locations to the positions of the three coils used in the MEG recordings, with the aid of digital photographs, to allow for an accurate co-registration of the two data sets. All anatomical landmarks digitized in the MEG study were made identifiable in the magnetic resonance images (MRI).

Data Analyses: VEFs were investigated via DataEditor (CTF Systems Inc.), a program that enables visualization of

MEG data. MEG data were averaged and DC-offset cor- rected with respect to the pre-stimulus baseline. All chan- nels were overlapped together for identification of evoked responses. The waveform was then visually inspected for the VEF components and eye blinks and/or artifacts. If a trial had noticeable eye blinks and/or artifacts, it was marked as bad and excluded from the averaging or source localization. The latency and amplitude of each component were then measured with DataEditor. We took the peak latency and amplitude of each component using overlapped waveforms instead of a single channel because the relative position between the subject's head and the MEG sensor arrays might vary across subjects. Frequency analysis was done with Morlet continuous wavelet transform. The fre- quency window for wavelet transform was 1-2000 Hz. The time window for wavelet transform was 0-600 ms. Our analysis focused on neuromagnetic activation in 40-400 ms following visual stimulation. Since wavelet transform is sensitive to time but not frequency at high-frequency sig- nals, a large sigma value (the number of waves) of Morlet continuous wavelet transform could enhance its sensitivity to frequency at high-frequency signals. On the contrary, wavelet transform is sensitive to frequency but not time in a low-frequency range. Therefore, a small sigma value of Morlet continuous wavelet transform could enhance its sensitivity to frequency at high-frequency signals. Building on this principle, we developed wave-cross method (see the following formula 1 and 2), which allowed us to set two sigma values.

Wc(t) = W(t,s1) x W(t,s2) (1)

W(t,s) = (2)

In the above formula, Wc(t) indicates the final results of wave-cross analysis based on two Morlet wavelet transforms: W(t,s1) and W(t,s2). Noteworthy, each wavelet transform has its own sigma value (s1 and s2). To produce one spectrogram for each time data set, a cross-product was computed for each time- frequency point with the coefficients corresponding to the two sigma values with the mother wavelet (2). If signals appeared in the giving sensitive time (a small sigma value) and sensitive frequency (a big sigma value) ranges, they would be enhanced. In the spectrogram, it looked like cross- ing one time-frequency point. Thus, the method was named as wave-cross analysis. Each trial of MEG data was trans- formed from time-domain waveform to frequency domain spectrogram individually with MEG Processor. Once all trials were transformed to spectral data; all trials were then accu- mulated one spectrogram.