Abstract
Purpose :
Most simulations of vision with scotomas are wrongly represented as black patches over the image. While paracentral scotomas are not explicitly observable by patients, central scotomas due to photoreceptor damage are reported as “blurred” vision. Pyramidal models representing the multiscale contrast processing of the visual system have been used to simulate vision with field loss. However, these simulations blended the low frequency image content, without completely removing it. Here, we present a physiologically plausible computation scheme for scene appearance with photoreceptor scotomas.
Methods :
After the pixel values in scotomatous areas in an image are replaced with “not a number” (NaN), we apply a pyramidal model (Peli, JOSA 1991, 2001) in the spatial domain. The interaction of the NaN with the pyramidal structure results in a simulated image that replaces the scotoma.
Results :
Content from outside the scotomatous area is drawn into the scotoma by the response of cells higher in the visual pathway centered within the scotoma but with wide receptive fields extending outside the scotoma. The resulting simulation with no black patch is consistent with patients’ descriptions (Figs 1 & 2). Due to the high spatial correlation typical of natural images, the replacement image is highly consistent with the elided part of the scene. The areas outside the scotomatous area are seen by the residual functioning normal peripheral low-resolution retina. The result is a filling-in like effect without any cortical processing or adaptation.
Conclusions :
The simulations provide insight into the lack of visibility of scotomas and further point to possible distinctions between field loss due to photoreceptor loss and higher cells loss (Peli, ARVO 2020).
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.