PatchMatch

Histopathology Image Retrieval System

PatchMatch is a content-based image retrieval system for digital pathology. Using deep learning and similarity search, it enables pathologists to find visually similar whole slide images (WSIs) from large archives, supporting diagnosis, research, and education.

Built with PyTorch, ResNet50, and FAISS, the system extracts features from tissue patches and performs fast similarity search across thousands of gigapixel pathology slides.

Team Members

Bertan Uran

Department of Computer Engineering, Bilkent University

Ekin Köylü

Department of Computer Engineering, Bilkent University

Elif Lara Oğuzhan

Department of Computer Engineering, Bilkent University

Emre Yazıcıoğlu

Department of Computer Engineering, Bilkent University

İlke Latifoğlu

Department of Computer Engineering, Bilkent University

Supervisor

Selim Aksoy, Ph.D.

Chair of the Department of Computer Engineering, Bilkent University

Industry Expert

Deniz Katırcıoğlu-Öztürk, Ph.D.

Medical AI Architect, SmartAlpha

View on GitHub →