Purchase this article with an account.
Eli Peli, Jae-Hyun Jung, Robert Goldstein; Better simulation of vision with central and paracentral scotomas. Invest. Ophthalmol. Vis. Sci. 2020;61(7):3370.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Simulations of vision with (para)central scotomas wrongly represent the scotoma as a black patch. Central scotomas are reported as “blurred” vision (actually low resolution). Paracentral scotomas are not explicitly observable by patients, though shrinking distortions were reported for cortical lesions. The ‘field loss’ is not apparent; it is missing from the patient’s consciousness and may be perceived only cognitively. Simulating such perceptions is a challenge. Better simulations are needed for training professionals, educating family members, caretakers, as a research tool for evaluation of vision rehabilitation, and for public advocacy.
We applied the Seam Carving technique (Avidan, ACM Graphics 2007) to remove the scotomatous area in an image resulting in the disappearance of contents while shrinking the image overall. Following seam adding operation the image is rescaled to the original size resulting in local distortions around the scotoma. We applied a Pyramidal Contrast Model with the eccentricity effect (Peli, JOSA 1991, 2001) to simulate the reduced resolution with eccentricity. We followed with the removal of the inputs for all layers (ganglion cells) of the pyramid centered at a central scotoma (Geisler, ACM 2002).
We obtained simulated perceived images (and videos) with no visible scotoma that are consistent in both cases with the description provided by patients with different causes for the field loss. Due to the larger receptive fields extending outside the central scotoma and the high spatial correlation in natural images, the scotoma area is illustrated as a very low resolution representation (Fig 1). The simulations provide insight into the lack of visibility of the scotoma as an imageless area in the scene and further point to possible distinctions between field loss due to photoreceptor loss vs. ganglion cells and cortex.
Novel simulations may provide insights. When implemented in real-time using a gaze-contingent display can provide a saccadic motion blur (Fig 2) in video augmented reality systems.
This is a 2020 ARVO Annual Meeting abstract.
Central scotoma simulated with pyramidal model. The scotoma (right) includes low resolution image on par with the far eccentricity to the left.
The effect of saccade shifting the seam carved paracentral scotoma (Avidan, 2007). On the bottom, a blurred version is inserted representing a saccadic suppression, masking distortions and induced motion (see https://youtu.be/OIGgPOb88jo).
This PDF is available to Subscribers Only