Is Complete Labeling Necessary? Understanding Active Learning in Longitudinal Medical Imaging

TL;DR

This paper introduces LMI-AL, a novel active learning framework for longitudinal medical imaging that efficiently reduces labeling effort while maintaining high detection performance of subtle changes.

Contribution

The study develops a specialized active learning method for change detection in longitudinal medical images, addressing the limitations of traditional DAL in this context.

Findings

Achieves comparable performance with less than 8% labeled data.

Demonstrates effectiveness of pairing and differencing slices for change detection.

Provides a publicly available codebase for future research.

Abstract

Detecting changes in longitudinal medical imaging using deep learning requires a substantial amount of accurately labeled data. However, labeling these images is notably more costly and time-consuming than labeling other image types, as it requires labeling across various time points, where new lesions can be minor, and subtle changes are easily missed. Deep Active Learning (DAL) has shown promise in minimizing labeling costs by selectively querying the most informative samples, but existing studies have primarily focused on static tasks like classification and segmentation. Consequently, the conventional DAL approach cannot be directly applied to change detection tasks, which involve identifying subtle differences across multiple images. In this study, we propose a novel DAL framework, named Longitudinal Medical Imaging Active Learning (LMI-AL), tailored specifically for longitudinal medical imaging. By pairing and differencing all 2D slices from baseline and follow-up 3D images, LMI-AL iteratively selects the most informative pairs for labeling using DAL, training a deep learning model with minimal manual annotation. Experimental results demonstrate that, with less than 8% of the data labeled, LMI-AL can achieve performance comparable to models trained on fully labeled datasets. We also provide a detailed analysis of the method's performance, as guidance for future research. The code is publicly available at https://github.com/HelenMa9998/Longitudinal_AL.