Explainable AI Framework for Enhanced Brain Tumour Detection and Classification

A brain tumour is rapid growth of cells in the brain. It needs accurate and early detection for effective treatment. This work introduces a novel approach to enhance interpretability of tumour. Using a publicly available dataset, this work focused on a baseline system using texture features and classified by Support Vector Machine (SVM) which resulted in an accuracy around $\mathbf{6 0 \%}$. Further, Deep Learning (DL) using finetuned Convolutional Neural Network (CNN) architecture is employed to improve system performance. This model achieves an accuracy of $\mathbf{9 8. 9 5 \%}$, sensitivity of $\mathbf{9 6. 8 1 \%}$, and specificity of $\mathbf{9 9. 6 5 \%}$. Explainable Artificial Intelligence (XAI) techniques like Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP) are also employed to improve the explainability of the detected tumour for enhancing the decision-making by radiologists.