MULTIMODAL EXPLAINABLE-AI FRAMEWORK FOR STROKE AND TRAUMATIC BRAIN INJURY DETECTION AND PROGNOSIS USING IMAGING AND CLINICAL DATA IN MACHINE LEARNING

Abstract

This research presents a Multimodal Explainable-AI Framework for the detection and prognosis of Stroke and Traumatic Brain Injury (TBI) using integrated imaging and clinical data. The proposed model leverages machine learning and deep learning techniques to enhance diagnostic accuracy and interpretability. Two datasets are employed: the Stroke Risk Prediction dataset containing patient clinical attributes and the TBI MRI Segmentation dataset providing brain imaging data. Preprocessing is performed using the Gaussian Wavelet Transform (GWT) to reduce noise and enhance feature clarity in MRI scans. Stroke and Traumatic Brain Injury (TBI) are employed for efficient feature extraction from multimodal data, capturing both spatial and textural patterns critical for diagnosis. The extracted features are classified using a Generalized Regression Neural Network (GRNN), which ensures fast training and robust generalization. To ensure transparency, Explainable-AI techniques are incorporated for interpretability of model decisions. Performance is evaluated using metrics such as accuracy of 0.91%, precision of 0.94%, recall of 0.90 %, F1-score of 0.91%, and ROC-AUC of 0.98%, demonstrating superior diagnostic capability compared to traditional models. This framework provides a powerful, interpretable decision-support system for clinicians, aiding early detection and prognosis assessment in stroke and TBI patients, ultimately contributing to improved treatment outcomes and patient.