Unified Model Agnostic Computation and Explainable AI for Enhanced Accuracy and Transparency in Medical Image Classification

The application of deep learning techniques in medical image classification has revolutionized healthcare diagnostics, but challenges remain in terms of model transparency and interpretability. Despite the high performance of deep learning models, their “black-box” nature often limits their clinical adoption, as healthcare professionals require an understanding of the rationale behind automated predictions. To address this, we propose a unified framework that integrates model-agnostic explainability techniques such as Shapley Additive Explanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) with deep learning models for medical image classification. This approach not only enhances the accuracy of the model by utilizing state-of-the-art convolutional neural network (CNN) architectures but also improves transparency by providing interpretable, humanunderstandable explanations for the model's decisions. The proposed framework is evaluated using various medical image datasets, including X-rays and MRIs, and is compared against traditional deep learning models without explainability methods. Results demonstrate that the integrated approach achieves superior classification accuracy while offering critical interpretability, making it more suitable for deployment in clinical settings. This work bridges the gap between highperformance deep learning models and the need for model transparency, promoting trust in AI-driven medical image analysis tools and enhancing their practical application in realworld healthcare scenarios.