Abstract
Human visual cortical areas have in the past been mapped with functional MRI by analyzing the phase-encoded responses to rotating wedge and expanding/contracting annular stimuli. We present here a method for the multifocal FMRI analysis of visual cortex, with the extraction of activation maps for multiple fixed visual field locations by decomposition of the compound response to test stimulus sequences applied concurrently to each location. The stimulus layout was a 60 region cortically-scaled dartboard of diameter 24o comprising 12 sectors in each of 5 concentric rings scaled to activate approximately equal areas in cortical area V1. For each 7.2 second block of the run, a different set of half of the 60 regions was active with 4×4 checkerboards reversing at 8 reversals/s, according to an orthogonal design. Two runs of 8 minutes give adequate results. Six subjects had scans done in a 3T GE Signa scanner, along with structural MRI and phase-encoded retinotopic analysis of conventional design. Data were analyzed with SPM2 with custom extensions, including optimization of the haemodynamic response model, noise autocorrelation and low-frequency trend removal. The resulting maps with these stimulus parameters reliably show each of the 60 regions activating area V1, in accordance with expected topography. Statistically significant activations also occurred in areas V2 and V3 for many regions, in the expected retinopic locations. The method demonstrates the viability of mapping human visual areas with FMRI using a multifocal array of 60 known visual field locations, and results are compared with maps from phase-encoded stimulation recorded in the same sessions. The method opens possibilities for more flexible design of what stimulus type may be presented at each location, allows estimation of spatial interactions, and also facilitates the use of identical spatial layouts for fMRI and in correlative visual evoked activity analysis using EEG and MEG recording.