Lesion Localization in OCT by Semi-Supervised Object Detection

TL;DR

This paper introduces a semi-supervised object detection approach for localizing retinal lesions in OCT images, utilizing a new dataset and identifying Unbiased Teacher as the best method, with improved accuracy over the baseline.

Contribution

It presents the first application of semi-supervised object detection to OCT lesion localization, including a new dataset and a structured taxonomy of SSOD methods.

Findings

Unbiased Teacher is the most effective SSOD method for OCT lesion localization.

The proposed method improves mAP from 49.34 to 50.86.

A new dataset with over 1k labeled and 13k unlabeled OCT images was created.

Abstract

Over 300 million people worldwide are affected by various retinal diseases. By noninvasive Optical Coherence Tomography (OCT) scans, a number of abnormal structural changes in the retina, namely retinal lesions, can be identified. Automated lesion localization in OCT is thus important for detecting retinal diseases at their early stage. To conquer the lack of manual annotation for deep supervised learning, this paper presents a first study on utilizing semi-supervised object detection (SSOD) for lesion localization in OCT images. To that end, we develop a taxonomy to provide a unified and structured viewpoint of the current SSOD methods, and consequently identify key modules in these methods. To evaluate the influence of these modules in the new task, we build OCT-SS, a new dataset consisting of over 1k expert-labeled OCT B-scan images and over 13k unlabeled B-scans. Extensive experiments on OCT-SS identify Unbiased Teacher (UnT) as the best current SSOD method for lesion localization. Moreover, we improve over this strong baseline, with mAP increased from 49.34 to 50.86.