MedScan — Deep Learning Diagnostics
CNN-based chest X-ray screening with explainable heatmap output for clinicians.
94.1%
Test accuracy
91%
Sensitivity
28k scans
Dataset size
The build
A research prototype for assisted radiology screening. We trained a convolutional classifier on a public chest X-ray corpus, added Grad-CAM heatmaps so a clinician can see what drove each prediction, and wrapped it in a review interface. Delivered with the training notebooks, evaluation metrics and a conference-ready paper draft.
This project was delivered under our FYP & Academic Projects service. Final year projects built properly — and explained properly.