Combined Spatial and Temporal Imaging of Form Discrimination in Humans---- A fMRI + ERP Study
Tiangang Zhou,Y. Zhuo,S. Fan,Lin Chen
2001-01-01
Abstract:Introduction Understanding the functional organization of brain structures underlying human perception and cognition requires knowledge of both the functional localizations of processing areas and the time dependence of processing. However, no single existing non-invasive neuroimaging technique, such as fMRI, PET or EEG, meets both the spatial and temporal resolution criteria[1~2]. In the present study, we combined fMRI and 128-channel EEG, in an effort for achieving millisecond temporal resolution and subcentimeter spatial resolution [3], to measure the spatiotemporal activation pattern of a shape discrimination task, which,as commonly believed, would be performed on the occipitotemporal pathway [4]. Methods The stimuli was that two identical triangles or disks were presented at left and right visual fields respectively, and subjects were asked to report what shape the stimulus was by pressing a corresponding one of two buttons. For fMRI scan, eight right-handed subjects were studied in a 1.5 T GE Signa scanner with a single-shot, T2*-weighted EPI sequence (GRE, TR/TE/θ=2000ms/40ms/60°, FOV: 25cm x 25cm, 5.0mm thk/2.5mm sp, 64x64) to acquire a set of 15 axial slices, obliquely aligned to parallel AC-PC line. 3 cycles were performed to obtain 48 (16x3) images for each stimulus in one slice. Data were analyzed offline with correlation method for each subject (r>0.4), and then compared across subjects to determine common activated areas. All data sets were transformed to Talairach coordinates and averaged across subjects to generate correlation coefficient maps using AFNI(R.Cox, U.Wisconsin). For ERP recordings, the same 8 subjects were recorded with a NeuroScan 128-channel EEG system in responding to the same task. For integration, both the fMRI-seeded dipole and moving dipole models were used to obtain time courses of the residual variance (RV) with time increment of 4ms using CURRY software. Results ERP results: The grand-average ERP waveforms at occipitotemporal sites (e.g. channel 102/108) (Fig.1A) showed a very distinct negative component N150 (peak latency 150 ms). The voltage topographic ERP maps of the N150 component within its time window (132-172 ms) (Fig.1B top), clearly showed bilateral posterior distribution. fMRI results: As shown in Fig.1B bottom, significant activation were found in inferior lateral occipitotemporal areas (averaged Talairach coordinates: x=17 y=57 z=-12). These activation areas, including the inferior temporal gyrus (GTi) and the fusiform gyrus (GF), corresponding to BA 19/37, are important in object and form perception [4]. Integration of fMRI & ERP: Inverse source analysis using different dipole models was performed on the grand-average and individual ERPs, respectively. Firstly, under fMRI-seeded dipoles model, using one pair of dipoles located at the averaged fMRI foci of GTi/GF, the RV value of the grand-average ERPs reached a best-fit (i.e.,lowest) point at 156 ms (RV=1.53%) within the time window of 132-172ms (mean RV=2.23%) (Fig.2A). For individual ERPs, the lowest RV occurred at a range of 132-176 ms (mean lowest point at 150ms, SD=16ms, mean RV=8.22%) (Fig.3). Secondly, under moving dipoles model, the RV value reached lowest at the same point as the fMRIseeded GTi/GF dipoles (for grand-average ERPs, lowest RV=1.42% at 156ms, mean RV=1.74% within time window of 132-172 ms; for individual ERPs, lowest RV occurred at a range of 132-176 ms, mean lowest point at 150 ms, SD=16ms, mean RV=5.34%). Furthermore, the position of the moving dipoles was near the fMRI foci of GTi/GF at the best-fit points (Fig.2B). These data provided additional evidence to the time course of the activation at GTi/GF in response to the task of form discrimination. Discussion The convergence between the results of fMRI and ERP recordings is evident in both the spatial and temporal respects. The first was the spatial convergence of the cortical activation in the occipitotemporal pathway. Well corresponding to the topographic map of N150, the fMRI signal enhancement was clearly observed in ventral occipitotemporal area, GTi/GF, during the task of form discrimination. These results were consistent with the findings in non-human primates as well as in humans [5~6] and supported the hypothesis of visual pathway that the ventral stream was responsible for form and object perception [4]. The second was the temporal convergence of the cortical activation. The best-fit latency 150 or 156 ms, from moving and seeded dipole models, was in a good agreement with the peak latency of the N1 component of ERPs. This data suggested that this time was actually the peak latency of the activation at ventral GTi/GF. Furthermore, the vicinity of the locations of moving dipoles and fMRI-seeded dipoles at the best-fit points strongly indicated that the activation at GTi/GF was actually the major source of the distinct negative component N1 of occipitotemporal ERPs. These converging results from the two imaging models, not only support the eligibility of each other, but also encouraged further studies of the integration of different imaging models to obtain the spatiotemporal characteristics of brain activation. References 1. Kwong KK et.al.. Proc.Nat.Acad.Sci. 89, 5675-5679 (1992) 2. Hamalainen MS et.al.. Rev.Mod.Phys. 65, 413-497 (1993) 3. Fox PT and Woldorff MG. Curr.Opin.Neurobiol. 4, 151-156 (1994). 4. Ungerleider LG and Haxby JV. Curr.Opin.Neurobiol. 4, 157-165 (1994). 5. Walsh V and Butler SR. Behaviour Brain Res. 76, 127-142 (1996). 6. Okusa T, Kakigi R and Osaka N. Neurosci. 98, 615-624 (2000)