msarvestani/sinusoidalTransform: V1
Madineh Sarvestani · Zenodo (CERN European Organization for Nuclear Research) · 2021
Supporting information for 'Sinusoidal transform of the visual field on the cortical surface', Sedigh-Sarvestani et al. 2021. This repo contains an interactive google Colab notebook for the map formation model in Figure 3. V2 is modeled, using an elastic-net, as consisting of 600 point ‘neurons’, each at a particular location on a cortical sheet and a receptive field (RF) defined by its azimuth and elevation in the visual field (VF). Retinotopic maps are produced by color-coding the sheet by the RF azimuth or elevation value of each neuron (Figure 3A, ‘Cortical Maps’). The premise is to create a topographic map by minimizing a cost function that trades off uniform visual field coverage with biologically inspired penalties applied to the cortical map. These penalties 'constrain' the mapping of the RFs according to their layout on the cortex and include: c1) smoothness c2) vertical meridian at area boundary and c3) relatively round RFs. Elastic nets implementation with tensorflow modified from Hsin-Hao Yu @ https://github.com/hsinhaoyu/DM_Retinotopy, who generously shared his code. To run this notebook: Runtime -> Run all. Alternatively you can run each cell by clicking the play button on the top left. Cells must be run in order. If you edit the code, the easiest way to ensure the edits get processed correctly is by restarting the notebook: Runtime -> Restart and run all. Madineh Sedigh-Sarvestani MPFI, 2021.