Reproduce “Comparative Evaluation of Deep Learning Models for Geographic Atrophy Segmentation in En Face OCT Images” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Utkarsh Doshi, Adarsh Gadari, Sharat Chandra Vupparaboina, Joseph DeCicco, Giulia Gregori, Nicola Valsecchi, Nasiq Hasan, Elham Sadeghi and 4 more
The authors present a standardized pipeline for segmenting geographic atrophy lesions in top-down projections from optical coherence tomography scans. They compare five learned models with Otsu thresholding on a fixed held-out benchmark from two devices. nnU-Net achieves the highest mean Dice score, 0.8882, but model tuning was not controlled, and area-agreement limits preclude fully automated individual progression monitoring.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.