郑州大学学报（理学版）2025，Vol.57Issue(5)：54-61,8.DOI:10.13705/j.issn.1671-6841.2023250
一种轻量级脑胶质瘤分割模型
A Lightweight Brain Glioma Segmentation Model
摘要
Abstract
In recent years,Transformer based automatic segmentation models for brain gliomas greatly improved in performance,but there were still some problems such as a large number of parameters,high computational power requirements,large sample requirements,and training difficulties,which could af-fect the actual deployment of the models.Therefore,a lightweight MRI glioma segmentation model MNATSPNet was proposed.Firstly,a lightweight component MobileNAT was designed to reduce the com-plexity of Transformer's multi-head self-attention through the adjacency attention mechanism.Secondly,the L1 structured pruning operation was introduced to remove the redundant parameters of the multi-head adjacency attention and feedforward neural network layer in MobileNAT.The experimental results demon-strated that MobileNAT and structured pruning operations could effectively reduce parameters of the model while maintaining stable segmentation performance.Finally,compared with other classic models,MNATSPNet achieved the best results.关键词
脑胶质瘤分割/注意力机制/Transformer/轻量化/结构化剪枝Key words
brain glioma segmentation/attention mechanism/Transformer/lightweight/structured pruning分类
信息技术与安全科学引用本文复制引用
呼伟,杨鸿杰,徐巧枝,智敏,萨和雅,于磊..一种轻量级脑胶质瘤分割模型[J].郑州大学学报（理学版）,2025,57(5):54-61,8.基金项目
国家自然科学基金项目(81660117) （81660117）
内蒙古自治区自然科学基金项目(2021MS06031) （2021MS06031）
内蒙古师范大学基本科研业务费专项(2022JBYJ034) （2022JBYJ034）
内蒙古自治区"十四五"社会公益领域重点研发和成果转化计划项目(2022YFSH0010) （2022YFSH0010）
无穷维哈密顿系统及其算法应用教育部重点实验室开放课题项目(2023KFYB06) （2023KFYB06）
