Histomancer: Liviu’s AI-Powered Slide Sorcery
A research prototype for real-time lung cancer subtype prediction using deep learning and histology slides.
Try the Diagnostic ToolRobust Classification
Trained on a combined dataset of real and augmented histopathology images, our EfficientNetB0-based model distinguishes between adenocarcinoma, squamous cell carcinoma, and benign tissue with over 98% test accuracy.
Real-Time Analysis
Upload histological images and receive AI-generated classifications instantly via our lightweight, browser-accessible web app.
Explainability Features
Saliency maps and Grad-CAM overlays reveal which regions of the image influence the AI's decisions, enhancing transparency and interpretability for researchers and medical professionals.