ExplainabilityTransparencyTrustworthinesscs.CV
Explainable AI based Glaucoma Detection using Transfer Learning and LIME
Touhidul Islam Chayan, Anita Islam, Eftykhar Rahman, Md. Tanzim Reza, Tasnim Sakib Apon, MD. Golam Rabiul Alam
October 7, 2022
2022 IEEE Asia-Pacific Conference on Computer Science and Data Engineering (CSDE)
22 citations
1 influential
Key Takeaways
-
AI can improve glaucoma detection accuracy while remaining understandable: The paper demonstrates that a transfer learning model achieved 94.71% accuracy in classifying glaucoma, while using LIME (Local Interpretable Model-Agnostic Explanations) to show doctors why the AI made each prediction, addressing a critical barrier to clinical adoption.
-
Explainability is essential for medical AI trust: The authors emphasize that traditional AI models are "black boxes" that prevent medical professionals from understanding predictions, making them unreliable for clinical decision-making—highlighting that high accuracy alone is insufficient for healthcare applications.
-
Explainable AI can reduce diagnostic errors: By providing doctors with "important and comprehensive information" about what visual features the AI identified, explainable systems help medical professionals make better-informed treatment decisions and catch cases that time-constrained clinicians might otherwise miss.
-
Transparency reduces AI fragility in healthcare: The paper argues that explainability "lessens the opacity and fragility of traditional deep learning models," suggesting that when doctors understand how an AI system works, they're better positioned to identify when it might fail or be misapplied.
Abstract
Glaucoma is the second driving reason for partial or complete blindness among all the visual deficiencies which mainly occurs because of excessive pressure in the eye due to anxiety or depression which damages the optic nerve and creates complications in vision. Traditional glaucoma screening is a time-consuming process that necessitates the medical professionals' constant attention, and even so time to time due to the time constrains and pressure they fail to classify correctly that leads to wrong treatment. Numerous efforts have been made to automate the entire glaucoma classification procedure however, these existing models in general have a black box characteristics that prevents users from understanding the key reasons behind the prediction and thus medical practitioners generally can not rely on these system. In this article after comparing with various pre-trained models, we propose a transfer learning model that is able to classify Glaucoma with 94.71\% accuracy. In addition, we have utilized Local Interpretable Model-Agnostic Explanations(LIME) that introduces explainability in our system. This improvement enables medical professionals obtain important and comprehensive information that aid them in making judgments. It also lessen the opacity and fragility of the traditional deep learning models.