Two-Stage Hybrid Supervision Framework for Fast, Low-resource, and Accurate Organ and Pan-cancer Segmentation in Abdomen CT

TL;DR

This paper introduces a two-stage hybrid supervision framework, StMt, for fast, low-resource, and accurate segmentation of abdominal organs and tumors in CT scans, leveraging partially labeled data and innovative input strategies.

Contribution

The paper presents a novel hybrid supervised framework combining self-training and mean teacher methods with a two-stage pipeline for improved segmentation accuracy and efficiency.

Findings

Achieved an average DSC of 89.79% for organs and 45.55% for lesions.

Demonstrated fast inference with an average of 11.25 seconds per case.

Utilized low GPU memory, with an average of 9627.82MB.

Abstract

Abdominal organ and tumour segmentation has many important clinical applications, such as organ quantification, surgical planning, and disease diagnosis. However, manual assessment is inherently subjective with considerable inter- and intra-expert variability. In the paper, we propose a hybrid supervised framework, StMt, that integrates self-training and mean teacher for the segmentation of abdominal organs and tumors using partially labeled and unlabeled data. We introduce a two-stage segmentation pipeline and whole-volume-based input strategy to maximize segmentation accuracy while meeting the requirements of inference time and GPU memory usage. Experiments on the validation set of FLARE2023 demonstrate that our method achieves excellent segmentation performance as well as fast and low-resource model inference. Our method achieved an average DSC score of 89.79\% and 45.55 \% for the organs and lesions on the validation set and the average running time and area under GPU memory-time cure are 11.25s and 9627.82MB, respectively.