Self-Supervised ImageNet Representations for In Vivo Confocal Microscopy: Tortuosity Grading without Segmentation Maps

TL;DR

This paper demonstrates that self-supervised ImageNet features, specifically DINO, can be fine-tuned to accurately grade corneal nerve fiber tortuosity in confocal microscopy images without segmentation maps.

Contribution

It shows the transferability of self-supervised features to medical imaging and improves tortuosity grading accuracy without segmentation maps.

Findings

DINO features outperform previous methods after fine-tuning.

Fine-tuned DINO achieves 84.25% accuracy and 77.97% sensitivity.

The approach eliminates the need for expensive segmentation maps.

Abstract

The tortuosity of corneal nerve fibers are used as indication for different diseases. Current state-of-the-art methods for grading the tortuosity heavily rely on expensive segmentation maps of these nerve fibers. In this paper, we demonstrate that self-supervised pretrained features from ImageNet are transferable to the domain of in vivo confocal microscopy. We show that DINO should not be disregarded as a deep learning model for medical imaging, although it was superseded by two later versions. After careful fine-tuning, DINO improves upon the state-of-the-art in terms of accuracy (84,25%) and sensitivity (77,97%). Our fine-tuned model focuses on the key morphological elements in grading without the use of segmentation maps.