Abstract
Recent advances in vision restoration techniques make it critical to objectively probe visual field (VF) defects, not only at the retinal but also at the cortical level. Here we assessed the potential of objective fMRI-based VF tests. To compare several fMRI-based VF-mapping approaches, we performed fMRI-based pRF-mapping (visual-field radius: 14°; 3T) in six patients with extensive VF-defects (4 glaucoma (GL); 2 retinitis pigmentosa (RP)) and six healthy controls (HC) with simulated peripheral scotoma (>7°). VF-coverage in V1 was reconstructed from (i) pRF-estimates[1], (ii) an anatomical retinotopic template[2], (iii) a Bayesian-inference approach[3] and compared to standard-automated perimetry (SAP) results. To assess the stimulus/task dependence, we also reconstructed VFs from fMRI data with block-design stimulation [drifting contrast patterns (8-directions) ON (12 s)/ OFF (12 s)] with different visual tasks, (i) passive viewing (PV) or (ii) a one-back task (OBT), i.e. reports of successions of identical motion directions. For the conventional-pRF-modeling approach, we found a stronger correspondence with SAP estimates in the controls [r=0.81 (median)] than in patients (r=0.57). The task-fMRI correlations were smaller than the pRF-mapping correlations, but followed a similar trend with the correspondence of patients’ falling short than controls. The differential correspondence for PV vs. OBT (r=0.29 vs. r=0.16) in patients, but not in controls (r=0.67 vs. r=0.65), indicated task-dependent dynamics in the VF-predictions in patients with VF-defects. The anatomy-based and Bayesian-based modeling approaches had better SAP-correspondence than the pRF approach in the patient group (PV: r=0.44 vs. r=0.42 vs. r=0.29; OBT: r=0.36 vs. r=0.32 vs. r=0.16). Our study demonstrates the feasibility fMRI-based VF-reconstructions, but also highlights the current limitations of translating fMRI-based methods to a clinical work-up. References: [1] Dumoulin & Wandell, 2008. https://doi.org/10.1016/j.neuroimage.2007.09.034 [2] Benson et al., 2014. https://doi.org/10.1371/journal.pcbi.1003538 [3] Benson & Winawer. 2018. https://doi.org/10.7554/eLife.40224