ConKeD++ -- Improving descriptor learning for retinal image registration: A comprehensive study of contrastive losses

TL;DR

This paper enhances a contrastive learning framework for retinal image registration, testing various loss functions, and demonstrating state-of-the-art results across multiple datasets, including new ones.

Contribution

It extends the ConKeD framework with improved contrastive losses and evaluates on new datasets, providing a comprehensive study and standardized evaluation approach.

Findings

Achieved state-of-the-art registration accuracy on all tested datasets.

Demonstrated advantages over existing methods in retinal image registration.

Released new datasets and evaluation protocols for future research.

Abstract

Self-supervised contrastive learning has emerged as one of the most successful deep learning paradigms. In this regard, it has seen extensive use in image registration and, more recently, in the particular field of medical image registration. In this work, we propose to test and extend and improve a state-of-the-art framework for color fundus image registration, ConKeD. Using the ConKeD framework we test multiple loss functions, adapting them to the framework and the application domain. Furthermore, we evaluate our models using the standarized benchmark dataset FIRE as well as several datasets that have never been used before for color fundus registration, for which we are releasing the pairing data as well as a standardized evaluation approach. Our work demonstrates state-of-the-art performance across all datasets and metrics demonstrating several advantages over current SOTA color fundus registration methods