Photo Credit: Sefa kart

Using an automated optical coherence tomography (OCT) segmentation algorithm to characterize and classify images from various inherited retinal disease (IRD) diagnostic classes, Pascal Escher, PhD, and colleagues showed the potential of artificial intelligence-assisted image processing to enhance IRD diagnostics and offer insights into the disease’s history. The study included 3184 images from 200 patients with IRD and 146 controls, including both healthy patients and with age-related macular degeneration. Automatic segmentation of six retinal layers and detection of nine AMD-related biomarkers was performed on retinal OCT images using an available AI tool. The researchers’ OCT-based diagnostic model reached 86% mean validation accuracy via four-fold cross-validation. Correlations between OCT features and genotypes revealed significant differences in gene-specific incidences of pathologic markers among 48 IRD genes. The researchers concluded that the specificity of their findings enables accurate classification of images, adding their findings could be further validated by expanding the analysis to additional IRD and control images.