计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (11A): 241100077-7.doi: 10.11896/jsjkx.241100077

• 计算机图形学&多媒体 • 上一篇    下一篇

基于全局-局部信息融合LPV-Net和3D-EDA的心脏半监督分割算法研究

胡慧称1, 刘瑞霞2, 刘照阳2, 郭振华3   

  1. 1 齐鲁工业大学(山东省科学院)数学与人工智能学部 济南 250353
    2 齐鲁工业大学(山东省科学院)山东省人工智能研究院 济南 250014
    3 浪潮电子信息产业股份有限公司 济南 250101
  • 出版日期:2025-11-15 发布日期:2025-11-10
  • 通讯作者: 刘瑞霞(liurx＠sdas.org)
  • 作者简介:15092704322@163.com
  • 基金资助:
    山东省自然科学基金创新发展联合基金项目(ZR2023LZH009)

Research on Heart Semi-supervised Segmentation Algorithm Based on Global-local Information Fusion LPV-Net and 3D-EDA

HU Huichen1, LIU Ruixia2, LIU Zhaoyang2, GUO Zhenhua3   

  1. 1 Department of Mathematics and Artificial Intelligence,Qilu University of Technology(Shandong Academy of Sciences),Jinan 250353,China
    2 Shandong Artificial Intelligence Institute,Qilu University of Technology(Shandong Academy of Sciences),Jinan 250014,China
    3 Inspur Electronic Information Industry Co.,Ltd.,Jinan 250101,China
  • Online:2025-11-15 Published:2025-11-10
  • Supported by:
    Shandong Provincial Natural Science Foundation Innovation and Development Joint Fund Project(ZR2023LZH009).

PDF (PC)

10

摘要： 心脏分割算法提供精准把握心脏结构的信息,辅助医生进行诊断、制定治疗计划以及进行手术前的评估工作,提高临床治疗的效果并减轻并发症。然而,心脏分割在应用中存在诸多问题。手动分割方法不仅耗时费力,并且具有很强的主观意识。全监督的心脏分割方法虽已取得一定成果,但标注数据的依赖性过高。现有的半监督分割方法在处理复杂的心脏结构和病理变化时表现仍不够理想,难以在实际临床环境中稳定应用。为解决这一问题,提出了心脏磁共振成像(MRI)分割方法,结合Linformer and Performer merge V-Net(LPV-Net)和3D Enhanced Discriminator with Attention(3D-EDA)技术,实现了全局-局部信息的有效整合。LPV-Net模块由LinPerBlock和改进的V-Net联袂打造,旨在规范模型训练过程、实现全局与局部信息的有机整合,有效提高分割效果的准确性与鲁棒性。同时引入新鉴别器3D-EDA规范未标记数据,并加入关键模块CARE-Layer,集成自定义注意力模块以增强对特征重要信息的捕捉能力,辅助网络可提高主网络分割指标性能。在左心房数据集上进行综合实验,并将所提方法与MC-Net,V-Net等其他先进的半监督方法进行比较,发现该方法在基准数据集上的表现尤为优异。特别是在使用有限标签数据进行训练时,该方法仍然展示出卓越的性能;当仅使用10%和20%的标记数据进行训练时,该方法的Dice系数分别达到88.50%和90.39%。

关键词: 半监督分割, 全局和局部信息, 融合, LPV-Net, 3D-EDA, 左心房, CARELayer

Abstract: Heart disease is one of the main causes of death worldwide,which seriously threatens human life and health.As a non-invasive medical imaging technology,cardiac magnetic resonance imaging(MRI) is widely used in clinical diagnosis,helping doctors accurately and efficiently diagnose and treat heart diseases.However,cardiac MRI segmentation faces great challenges in practical applications,manual segmentation methods are time-consuming and subjective,while existing fully supervised and semi-supervised segmentation methods are not effective in dealing with complex cardiac structures and pathological changes,limited by the scarcity of data sets.This study aims to solve the challenges of cardiac MRI segmentation by proposing a 3D left atrial semi-supervised segmentation framework based on global-local information fusion to address the time-consuming and subjective pro-blems of manual segmentation.Although the current fully supervised heart segmentation methods are effective,they are limited by the scarcity of data sets.The semi-supervised methods come into being,but they are still limited by the small amount of data,especially when dealing with complex cardiac structure and pathological changes.To solve this problem,this study propose a new cardiac MRI segmentation method that combines Linformer and Performer merge V-Net(LPV-Net) and 3D Enhanced Discriminator with Attention(3D-EDA) technologies to achieve an effective fusion of global-local information.The LPV-Net module,created by LinPerBlock and improved V-Net,aims to standardize the training process of the model and achieve effective fusion of global and local information,thus improving the accuracy and robustness of segmentation.In addition,we also introduced a new discriminator 3D-EDA for the specification of unlabeled data.The most critical module in the model is CARELayer,which integrates a custom attention module to enhance the ability of the model to capture important information in the feature,and the auxiliary segmentation network improves the segmentation performance.By conducting a comprehensive experiment on the left atrial dataset,comparing the proposed method with several advanced semi-supervised methods.The experimental results show that the proposed method performs well on the baseline dataset,especially when training with limited label data.For example,when training with only 10% and 20% labeled data,the Dice coefficients of 88.50％ and 90.39％ were obtained.

Key words: Semi-supervised segmentation, Global and local information, Fuse, LPV-Net, 3D-EDA, Left atrium, CARELayer

中图分类号: 

  • TP391
[1]ZHANG Y,JI Y.HINT1(Histidine Triad Nucleotide-Binding Protein 1) Attenuates Cardiac Hypertrophy Via Suppressing HOXA5(Homeobox A5) Expression[J].Circulation,2022,145(8):e151-e152.
[2]JIANG L S,HAO Z Y,XIE X Y,et al.Left atrial appendage angiography for stroke risk prediction in patients with atrial fibrillation[J].Eurointervention,2023:19(8):695-702.
[3]GASSENMAIER S.Deep learning applications in magnetic resonance imaging:has the future become present?[J].Diagnostics,2021,11(12):2181.
[4]OLYA,HESSAM M.An integrated deep learning and stochastic optimization approach for resource management in team-based healthcare systems[J].Expert Systems with Applications,2022(187):115924.
[5]XIONG Z,FEDOROV V V,FU X,et al.Fully automatic leftatrium segmentation from late gadolinium enhanced magnetic resonance imaging using a dual fully convolutional neural network[J].IEEE Transactions on Medical Imaging,2019,38(2):515-524.
[6]XIA Q,YAO Y,HU Z,et al.Automatic 3D atrial segmentation from GE-MRIs using volumetric fully convolutional networks[C]//International Workshop on Statistical Atlases and Computational Models of the Heart.2018:211-220.
[7]LI L,WENG X,SCHNABEL J A,et al.Joint left atrial segmentation and scar quantification based on a DNN with spatial encoding and shape attention[C]//Medical Image Computing and Computer Assisted Intervention(MICCAI).2020:118-127.
[8]ZHANG H,ZHANG W,SHEN W,et al.Automatic segmentation of the cardiac MR images based on nested fully convolutional dense network with dilated convolution[J].Biomedical Signal Processing and Control,2021,68:102684.
[9]CHEN C,BAI W,RUECKERT D.Multi-task learning for left atrial segmentation on GE-MRI[C]//Statistical Atlases and Computational Models of the Heart.2018:292-301.
[10]XIONG Z,XIA Q,HU Z,et al.A global benchmark of algorithms for segmenting the left atrium from late gadolinium-enhanced cardiac magnetic resonance imaging[J].Medical Image Analysis,2021,67:101832.
[11]RODRIGUEZ A,ALESSANDRO L.Clustering by fast search and find of density peaks[J].Science,2014,344(6191):1492-1496.
[12]GOODFELLOW I.Generative adversarial networks[J].Communications of the ACM,2020,63(11):139-144.
[13]MIRZA M,SIMON O.Conditional generative adversarial nets[J].arXiv:1411.1784,2014.
[14]BORTSOVA G,DUBOST F,HOGEWEG L,et al.Semi-supervised medical image segmentation via learning consistency under transformations[C]//Medical Image Computing and Computer Assisted Intervention(MICCAI).2019:810-818.
[15]YU L,WANG S,LI X,et al.Uncertainty-aware self-ensembling model for semi-supervised 3D left atrium segmentation[C]//Medical Image Computing and Computer Assisted Intervention(MICCAI).2019:605-613.
[16]LUO X,CHEN J,SONG T.Semi-supervised medical image segmentation through dual-task consistency[C]//Proceeding of the AAAI Conference on Artificial Intelligence(AAAI).2021:8801-8809.
[17]LI S,ZHANG C,HE X.Shape-aware semi-supervised 3D semantic segmentation for medical images[C]//Medical Image Computing and Computer Assisted Intervention(MICCAI).2020:552-561.
[18]WU Y,XU M,GE Z,et al.Semi-supervised left atrium segmen-tation with mutual consistency training[C]//MICCAI 2021.2021:297-306.
[19]CHEN X,YUAN Y,ZENG G,et al.Semi-supervised semantic segmentation with cross pseudo supervision[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2021:2613-2622.
[20]MA J,WEI Z,ZHANG Y,et al.How distance transform maps boost segmentation CNNs:an empirical study[C]//Medical Imaging with Deep Learning.PMLR,2020:479-492.
[21]OKTAY O.Attention u-net:Learning where to look for the pancreas[J].arXiv:1804.03999,2018.
[22]WANG S N,WEI B Z,DNLE S,et al.Linformer:Self-attention with linear complexity[J].arXiv:2006.04768,2020.
[23]CHOROMANSKI K.Rethinking attention with performers[J].arXiv:2009.14794,2020.
[24]MILLETARI F,NASSIR N,SEYED-AHMAD A.V-net:Fully convolutional neural networks for volumetric medical image segmentation[C]//2016 Fourth International Conference on 3D Vision(3DV).IEEE,2016:565-571.
[25]PARK J J,FLORENCE P,STRAUB J,et al.Deepsdf:Learning continuous signed distance functions for shape representation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2019:165-174.
[26]DICE L R.Measures of the amount of ecologic association between species[J].Ecology,1945,26(3):297-302.
[27]ZHANG K.A comprehensive review on medical image segmentation[J].Computers in Biology and Medicine,2017,91:11-29.
[28]BERTHELOT D.BEGAN:Boundary equilibrium generative adversarial networks[J].International Conference on Learning Representations(ICLR).arXiv:1703.10717,2017.
[29]GOODFELLOW I J,JEAN P A,MEHDI M,et al.Generativeadversarial nets[J].NeurIPS 2014,2014,2:2672-2680.[30]LUO X,WANG G,LIAO W,et al.Semi-supervised medical image segmentation via uncertainty rectified pyramid consistency[J].Medical Image Analysis,2022,80:102517.
[31]WU Y,WU Z,WU Q,et al.Exploring smoothness and class-separation for semi-supervised medical image segmentation[C]//25th International Conference Medical Image Computing and Computer Assisted Intervention(MICCAI 2022).2022:34-43.
[32]WANG Y,ZHANG Y,TIAN J,et al.Doubleuncertainty weighted method for semi-supervised learning[C]//International Conference on Medical Image Computing and Computer-Assisted Intervention.2020:542-551.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发