Meeting Banner
Abstract #1864

A Comprehensive Deep Learning Approach for Multi-type Central Nervous System Tumor Segmentation Based on the 2021 WHO Classification

Shuang Li1, Liting Guo2, Yuchi Tian 3, Guangliang Ju2, Simin Zhang 1, Yuqi Jin1, Xiaorui Su1, Shuang Tang1, Anrong Zeng1, Yuting Luo1, Xibiao Yang1, Lang Wang1,4, Liping Wang1, Hongjing Zhang4, Weilin Yang1,5, Xiaoyun Liang3, and Qiang Yue1
1West China Hospital of Sichuan University, Chengdu, China, 2Intelligent Imaging Software R&D Division, Neusoft Medical Systems Co., Ltd, Shenyang, China, 3Institute of Research and Clinical Innovations,Neusoft Medical Systems Co., Ltd, Shanghai, China, 4West Chian fourth Hospital of Sichuan University, Chengdu, China, 5West China Second University Hospital, Chengdu, China

Synopsis

Keywords: Other AI/ML, Machine Learning/Artificial Intelligence, Central Nervous System; Tumor ;segmentation

Motivation: CNS tumor segmentation is essential but remains limited by time-intensive methods and variability. Current automated approaches typically focus on single tumor types, lacking comprehensive clinical utility.

Goal(s): Develop a unified deep learning model for accurate segmentation of 11 CNS tumor types using standard MRI sequences, addressing the limitations of existing single-tumor models.

Approach: A 3D ResNet-Transformer framework was employed for feature extraction and attention-based enhancement. The model was trained and evaluated on a dataset combining public and hospital-sourced MRI scans.

Results: The model achieved mean Dice scores of 0.896 demonstrating robust multi-tumor segmentation, though variability was noted in rarer types.

Impact: This work advances automated, multi-tumor segmentation tools, enhancing clinical workflows and supporting consistent, efficient CNS tumor analysis.

How to access this content:

For one year after publication, abstracts and videos are only open to registrants of this annual meeting. Registrants should use their existing login information. Non-registrant access can be purchased via the ISMRM E-Library.

After one year, current ISMRM & ISMRT members get free access to both the abstracts and videos. Non-members and non-registrants must purchase access via the ISMRM E-Library.

After two years, the meeting proceedings (abstracts) are opened to the public and require no login information. Videos remain behind password for access by members, registrants and E-Library customers.

Click here for more information on becoming a member.

Keywords