Medical Image Classification for Pneumonia & Brain Tumor Detection
Clinical-grade deep transfer learning model with Grad-CAM visual heatmaps, classifying Chest X-Rays and Brain MRI scans with 97.8% diagnostic validation.
Detailed Project Overview
An explainable AI medical diagnosis platform built with PyTorch and DenseNet-121 / EfficientNet-B4. The system ingests DICOM/JPEG chest radiographs and MRI scans, pre-processes them with adaptive histogram equalization (CLAHE), and highlights anomalous tissue regions using Gradient-weighted Class Activation Mapping (Grad-CAM). Includes automated PDF radiology report synthesis.
Academic Problem Statement
Radiologists in rural and under-resourced hospitals face heavy backlogs, leading to delayed diagnoses of critical pulmonary infections and cerebral lesions. An explainable decision-support tool is essential to assist clinicians without acting as an uninterpretable "black box".
Proposed Methodology & Flow
More Projects in Artificial Intelligence & Machine Learning
Explore alternative architectures and applications in this discipline.
AI-Powered Smart Attendance & Anti-Spoofing Facial Recognition
Real-time multi-face attendance logging with liveness anti-spoofing detection using YOLOv8, FaceNet embeddings, and automated faculty analytics dashboard.
Autonomous Vehicle ADAS Lane & Obstacle Collision Warning System
Advanced Driver Assistance System with dynamic polynomial lane curve estimation, YOLOv8 vehicle/pedestrian distance bounding, and Time-to-Collision (TTC) acoustic warning.
Real-Time Misinformation & Fake News Detector with Explainable BERT
Deep bidirectional transformer NLP model evaluating news credibility, emotional sensationalism bias, and source provenance with Chrome Extension integration.