Negar Jalili Mallak¹ (), Zhong-Lin Lu^2,3,4, Yalin Wang¹; ¹School of Computing, Informatics, and Decision Systems Engineering, Arizona State University, Tempe, AZ, USA., ²Division of Arts and Sciences, NYU Shanghai, Shanghai, China., ³Center for Neural Science and Department of Psychology, New York University, New York, USA., ⁴NYU-ECNU Institute of Brain and Cognitive Science, NYU Shanghai, Shanghai, China.

Retinotopic mapping, a fundamental component of visual cognitive neuroscience, helps us understand how the brain processes visual stimuli. Despite its significance, the process relies on BOLD functional magnetic resonance imaging (fMRI), which has a low signal–noise ratio (SNR) and low spatial resolution. These limitations impede the creation of accurate and precise retinotopic maps. This study introduces a novel application of Diffeomorphic Registration for Retinotopic Maps (DRRM; Tu, et al, 2022) to enhance the alignment of retinotopic maps using the 3T NYU Retinotopy Dataset, encompassing analyze-PRF and mrVista results. Diffeomorphic Registration for Retinotopic Maps (DRRM) quantifies the diffeomorphic condition, ensuring accurate alignment of retinotopic maps under topological conditions by leveraging the Beltrami coefficient. We evaluated the quality of the registered retinotopic maps utilizing visual coordinate change, flipped triangles, and goodness of fit to BOLD time series. Minimized visual coordinate changes and the absence of flipped triangles validate the diffeomorphic nature of DRRM registration. The results indicate that the DRRM-registered retinotopic maps provide a superior fit to the fMRI time series which is evident in the reduced Root Mean Square Error (RMSE) from the DRRM fits (mean RMSE=1.06) compared to the structurally registered retinotopic maps and Benson’s inference map (mean RMSE= 1.3). This improvement suggests that visual coordinates from the DRRM provided a better account of the fMRI time series than the original population receptive field solutions. In conclusion, our work demonstrates that DRRM is a valuable tool that can significantly improve the quality of retinotopic maps in the realm of 3T fMRI data. This successful adaptation positions DRRM as a promising method for advancing retinotopic map research and applications, addressing the challenges posed by the limited SNR and spatial resolution of BOLD fMRI and enhancing the accuracy and interpretability of retinotopic maps.

Acknowledgements: The work was partially supported by NSF (DMS-1413417 and DMS-1412722) and NIH (R01EY032125 and R01DE030286)