EXPLAINABLE DEEP LEARNING FOR TRUSTWORTHY MEDICAL IMAGE DIAGNOSIS: A HYBRID AI FRAMEWORK

Abstract

Deep Learning has enhanced the efficacy of Automated Medical Image Diagnosis and Disease Classification. The non-Transparency and non-Interpretability of these models restrict the applicability of their automated diagnostic systems. This study presents an Explainable Deep Learning for Trustworthy Medical Image Diagnosis. This Hybrid AI Framework utilizes modern Deep Learning and Artificial Intelligence (AI) to develop a more accurate diagnosis with greater transparency. This framework integrates Deep Learning models for feature extraction with hybrid AI to facilitate the generation of precise diagnostic predictions and to concomitantly offer diagnostic rationales through visual and interpretive feature Analyses. This work is interested in enhancing the Trustworthiness of AI solutions for automated Medical Diagnosis. This is accomplished by striving to remedy the non-Interpretability, non-Trust, and uncertainty of framework’s models. The framework’s Trustworthiness, Interpretability, and Explainability are assessed to evaluate the framework’s diagnostic accuracy and its applicability in Healthcare. This work is an attempt to develop imaging systems for the automated diagnosis of medical imaging that are more transparent, easier to use, and more trustworthy by bridging the gap between effective AI systems and the processes of clinical decision making