2019/06/04 by Thomas W. Rogers, Rogers, Thomas W., Nicolas Jaccard +11
Medicine · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Glaucoma and retinal disorders #Image and Video Processing (eess.IV) #Retinal Diseases and Treatments #Retinal Imaging and Analysis #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1906.01272
openalex publication_date 2019/06/04 · openalex created_date 2022/07/29 · openalex updated_date 2026/07/28
Objectives: To evaluate the performance of a deep learning based Artificial\nIntelligence (AI) software for detection of glaucoma from stereoscopic optic\ndisc photographs, and to compare this performance to the performance of a large\ncohort of ophthalmologists and optometrists.\n Methods: A retrospective study evaluating the diagnostic performance of an AI\nsoftware (Pegasus v1.0, Visulytix Ltd., London UK) and comparing it to that of\n243 European ophthalmologists and 208 British optometrists, as determined in\nprevious studies, for the detection of glaucomatous optic neuropathy from 94\nscanned stereoscopic photographic slides scanned into digital format.\n Results: Pegasus was able to detect glaucomatous optic neuropathy with an\naccuracy of 83.4% (95% CI: 77.5-89.2). This is comparable to an average\nophthalmologist accuracy of 80.5% (95% CI: 67.2-93.8) and average optometrist\naccuracy of 80% (95% CI: 67-88) on the same images. In addition, the AI system\nhad an intra-observer agreement (Cohen's Kappa, \κ) of 0.74 (95% CI:\n0.63-0.85), compared to 0.70 (range: -0.13-1.00; 95% CI: 0.67-0.73) and 0.71\n(range: 0.08-1.00) for ophthalmologists and optometrists, respectively. There\nwas no statistically significant difference between the performance of the deep\nlearning system and ophthalmologists or optometrists. There was no\nstatistically significant difference between the performance of the deep\nlearning system and ophthalmologists or optometrists.\n Conclusion: The AI system obtained a diagnostic performance and repeatability\ncomparable to that of the ophthalmologists and optometrists. We conclude that\ndeep learning based AI systems, such as Pegasus, demonstrate significant\npromise in the assisted detection of glaucomatous optic neuropathy.\n