基于个性化联邦学习的多源域MRI左心室分割  

Multi-Source Domain MRI Deft Ventricular Segmentation Based on Personalized Federated Learning

作　　者：李纯真 LI Chunzhen(School of advanced manufacturing,Fuzhou University,Quanzhou 362200,China)

机构地区：[1]福州大学先进制造学院,福建泉州362200

出　　处：《电视技术》2023年第8期16-19,42,共5页Video Engineering

摘　　要：心脏分割是医学影像分析的一个重要任务,可以用于诊断和评估各种心血管疾病。然而,利用多源域的心脏磁共振成像(Magnetic Resonance Imaging,MRI)数据进行联邦学习建模,仍然是一个挑战。针对当前数据隐私和多源域数据分布异构的问题,将个性化联邦学习与直方图匹配结合,在不共享本地数据的情况下,联合多个医疗机构进行分布式学习,减轻多个源域数据之间的数据差异,提高模型准确率。实验结果表明,所提出的个性化联邦学习方法在多源域的心脏MRI图像左心室分割任务上达到了较高的准确率。Cardiac segmentation is an important task in medical image analysis,which can be used to diagnose and evaluate various cardiovascular diseases.However,the use of multi-source cardiac Magnetic Resonance Imaging(MRI)data for federated learning modeling remains a challenge.Aiming at the current problems of data privacy and heterogeneous distribution of multi-source domain data,personalized federated learning is combined with histogram matching.Without sharing local data,multiple medical institutions are combined for distributed learning to reduce data differences between multiple source domain data and improve model accuracy.The experimental results show that the proposed personalized federated learning method achieves high accuracy in the multi-source domain cardiac MRI image left ventricular segmentation task.

关 键 词：磁共振成像(MRI) 联邦学习 直方图匹配 医疗图像分割 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]