Learning to Segment using Summary Statistics and Weak Supervision

TL;DR

This paper introduces a method for training image segmentation models using only summary statistics and minimal weak supervision, reducing the need for detailed annotations.

Contribution

The authors propose a novel loss function that combines image reconstruction, summary statistic matching, and weak pixel supervision to improve segmentation accuracy.

Findings

Adding weak pixel supervision significantly enhances segmentation performance.

The approach works effectively across standard images, ultrasound, and CT scan data.

Abstract

Medical experts often manually segment images to obtain diagnostic statistics and discard the resulting annotations. We aim to train segmentation models to alleviate this burden, but constrained to the retained summary statistics (e.g., the area of the annotated region). Empirical results suggest that statistics alone are insufficient for this task, but adding weak information in the form of a few pixels within the area of interest significantly improves performance. We use a novel loss function that combines terms for image reconstruction quality, matching to summary statistics, and overlap between the predicted foreground and the weak supervisory signal. Experiments on standard image, ultrasound (breast cancer), and Computed Tomography (CT) scan (kidney tumors) data demonstrate the utility and potential of the approach.