MSWAL: 3D Multi-class Segmentation of Whole Abdominal Lesions Dataset

2025-03-17

AI Summary

Existing abdominal lesion segmentation models are hindered by scarce annotated data and limited disease coverage. To address this, we introduce MSWAL—the first large-scale, fully annotated 3D CT dataset for comprehensive abdominal multi-disease segmentation—covering seven lesion types: gallstones, renal calculi, and hepatic, renal, and pancreatic tumors and cysts. MSWAL comprises 694 multi-phase CT scans (191,417 axial slices), with deliberate diversity in patient sex, imaging equipment, and acquisition protocols. We propose a novel fine-grained, whole-abdomen 3D multi-class annotation paradigm and design Inception nnU-Net, which integrates multi-scale receptive field features via inception-style convolutions. On MSWAL, our method achieves significant improvements in voxel-wise Dice (+2.15%) and region-level F1 (+3.42%). When transferred to LiTS and KiTS benchmarks, it boosts liver and kidney tumor Dice scores by 3.00% and 0.89%, respectively. Code is publicly available; the MSWAL dataset will be released upon publication.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

With the significantly increasing incidence and prevalence of abdominal diseases, there is a need to embrace greater use of new innovations and technology for the diagnosis and treatment of patients. Although deep-learning methods have notably been developed to assist radiologists in diagnosing abdominal diseases, existing models have the restricted ability to segment common lesions in the abdomen due to missing annotations for typical abdominal pathologies in their training datasets. To address the limitation, we introduce MSWAL, the first 3D Multi-class Segmentation of the Whole Abdominal Lesions dataset, which broadens the coverage of various common lesion types, such as gallstones, kidney stones, liver tumors, kidney tumors, pancreatic cancer, liver cysts, and kidney cysts. With CT scans collected from 694 patients (191,417 slices) of different genders across various scanning phases, MSWAL demonstrates strong robustness and generalizability. The transfer learning experiment from MSWAL to two public datasets, LiTS and KiTS, effectively demonstrates consistent improvements, with Dice Similarity Coefficient (DSC) increase of 3.00% for liver tumors and 0.89% for kidney tumors, demonstrating that the comprehensive annotations and diverse lesion types in MSWAL facilitate effective learning across different domains and data distributions. Furthermore, we propose Inception nnU-Net, a novel segmentation framework that effectively integrates an Inception module with the nnU-Net architecture to extract information from different receptive fields, achieving significant enhancement in both voxel-level DSC and region-level F1 compared to the cutting-edge public algorithms on MSWAL. Our dataset will be released after being accepted, and the code is publicly released at https://github.com/tiuxuxsh76075/MSWAL-.

Problem

• Limited lesion segmentation due to missing annotations
• Need for robust 3D multi-class abdominal lesion dataset
• Improving segmentation accuracy across diverse lesion types

Innovation

• MSWAL: 3D multi-class abdominal lesion segmentation
• Inception nnU-Net: enhanced segmentation framework
• Transfer learning improves liver and kidney tumor segmentation