[ Case Study ]
TBC Detector
AI-powered tuberculosis screening from chest X-rays using EfficientNetB0 transfer learning with a full-stack web interface.
Trained an EfficientNetB0 transfer learning model on chest X-ray data with threshold tuning for high sensitivity, served via FastAPI with a React frontend and Docker Compose orchestration.
Core Problem
Tuberculosis screening from chest X-rays requires expert radiologist review, which is slow and resource-intensive in under-resourced settings.
Solution
Trained an EfficientNetB0 transfer learning model on chest X-ray data with threshold tuning for high sensitivity, served via FastAPI with a React frontend and Docker Compose orchestration.
Outcome
Achieved approximately 98.4% validation accuracy. Delivered a complete academic project demonstrating applied deep learning in a medical screening context.
Selected screens
