Explainable AI in Medical Imaging: A Hybrid CNN and Random Forest Framework for Brain Tumor Detection

A critical disease, a brain tumor must be categorized correctly if treatment and patient management are to be directed. Due to interpretational bias by individuals and errors that are rife in human practice, traditional diagnostic methods sometimes prove wanting. To increase the accuracy of differentiating brain tumors from similar tissues in MR images, this work explores modern computational approaches-particularly CNNs and RFs. This is evidenced by reviewing related literature on MRI-based brain tumor detection, for which deep learning methods are deemed efficient but challenging. The proposed strategy entails preprocessing of the MRI then using CNNs in feature extraction and enhancing performance by the Random Forest classifier. Further, the work uses Explainable Artificial Intelligence (X-AI) approaches, specifically Local Interpretable Model-agnostic Explanations (LIME), to address the interpretability challenge common in healthcare. Meningiomas yield a 98.5% accuracy, and the other performance parameters calculated that depict a high degree of precision and recall with many tumor types suggest a reasonable classification. This twin objective of maintaining accuracy and interpretability not only enhances the decision-making process of clinicians but also assures trust in deploying AI solutions for producing better patient outcomes in the case of brain tumors.