ULS+: Data-driven Model Adaptation Enhances Lesion Segmentation

TL;DR

ULS+ is an improved lesion segmentation model that integrates new datasets and optimizes input sizes, achieving higher accuracy and faster inference, and leading on the ULS23 Challenge leaderboard.

Contribution

This paper introduces ULS+, a data-driven enhancement of the ULS model that incorporates additional datasets and smaller inputs for improved performance.

Findings

ULS+ outperforms ULS in Dice score and robustness.

ULS+ ranks first on the ULS23 Challenge leaderboard.

ULS+ demonstrates higher accuracy and faster inference.

Abstract

In this study, we present ULS+, an enhanced version of the Universal Lesion Segmentation (ULS) model. The original ULS model segments lesions across the whole body in CT scans given volumes of interest (VOIs) centered around a click-point. Since its release, several new public datasets have become available that can further improve model performance. ULS+ incorporates these additional datasets and uses smaller input image sizes, resulting in higher accuracy and faster inference. We compared ULS and ULS+ using the Dice score and robustness to click-point location on the ULS23 Challenge test data and a subset of the Longitudinal-CT dataset. In all comparisons, ULS+ significantly outperformed ULS. Additionally, ULS+ ranks first on the ULS23 Challenge test-phase leaderboard. By maintaining a cycle of data-driven updates and clinical validation, ULS+ establishes a foundation for robust and clinically relevant lesion segmentation models.