Enhanced Medical Imaging –A Multimodal Explainable AI-Driven X-Ray Diagnosis

Medical image analysis, particularly Xray interpretation, is a critical challenge in modern healthcare due to the global shortage of trained radiologists and the increasing volume of diagnostic imaging. This paper presents an AIPowered Multi-Disease X-Ray Analysis System that automates body part identification and pathological disease detection from X-ray images using deep learning. The proposed system employs a hybrid body part detection module combining Contrastive Language–Image Pre-training (CLIP) and Convolutional Neural Networks (CNN), achieving robust classification across 16 anatomical categories. Disease detection is performed using a DenseNet-121 model pretrained on ChestX-ray14 for 10 chest pathologies and a DenseNet-169 model fine-tuned on the MURA dataset for musculoskeletal conditions. An integrated Explainable AI (XAI) module provides Gradient-weighted Class Activation Mapping (Grad-CAM) heatmaps, saliency maps, feature importance visualization, and confidence breakdowns to enhance diagnostic transparency. The system generates structured multilingual clinical reports in English and Telugu with text-tospeech voice narration via gTTS. Patient records and analysis histories are persistently managed through a MongoDB Atlas cloud database with GridFS PDF storage. Deployed as an interactive Gradio web application, the system achieves a mean body part detection accuracy and an average AUC-ROC for chest pathology detection, demonstrating strong clinical utility as an AIassisted diagnostic support tool.