arXiv 2502.167092025-02-23eess.IVcs.CV

FedDA-TSformer：基于门控心肌灌注SPECT图像的左心室分割的联邦域适应与视觉TimeSformer

FedDA-TSformer: Federated Domain Adaptation with Vision TimeSformer for Left Ventricle Segmentation on Gated Myocardial Perfusion SPECT Image

Kennesaw State University（肯尼索州立大学）；Michigan Technological University（密歇根理工大学）；Xiangya Hospital, Central South University（中南大学湘雅医院）；Chang Bing Show Chwan Memorial Hospital（彰滨秀传纪念医院）；The First Affiliated Hospital of Nanjing Medical University（南京医科大学第一附属医院）；Health Research Institute, Michigan Technological University（密歇根理工大学健康研究所）

Yehong Huang, Chen Zhao, Rochak Dhakal, Min Zhao, Guang-Uei Hung, Zhixin Jiang, Weihua Zhou

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出FedDA-TSformer，融合联邦域适应与TimeSformer，利用时空注意力及LMMD损失，实现多中心门控心肌灌注图像左心室分割，兼顾隐私与性能。

摘要：

背景与目的：利用门控心肌灌注（MPS）单光子发射计算机断层扫描进行左心室功能评估，依赖于精确提取左心室轮廓，同时确保患者数据的安全性。方法：本文介绍了将联邦域适应与TimeSformer相结合的方法，命名为“FedDA-TSformer”，用于基于MPS的左心室分割。FedDA-TSformer捕获门控MPS图像中的空间和时间特征，利用空间注意力、时间注意力和联邦学习来改进域适应，同时确保患者数据安全。具体而言，我们采用了Divide-Space-Time-Attention机制，从多中心MPS数据集中提取时空相关性，确保预测在时空上保持一致。为了实现域适应，我们使用局部最大均值差异（LMMD）损失对齐模型在来自三个不同中心的MPS上的输出。该方法有效解决了联邦学习和域适应的双重需求，增强了模型在多站点数据集训练过程中的性能，同时确保来自不同医院的数据得到保护。结果：我们的FedDA-TSformer使用来自三家医院收集的MPS数据集进行训练和评估，共包含150名受试者。每位受试者的心动周期被分为八个门控。模型在左心室（LV）心内膜和心外膜分割上分别获得了Dice相似系数（DSC）0.842和0.907。结论：我们提出的FedDA-TSformer模型解决了多中心泛化的挑战，确保了患者数据隐私保护，并在左心室（LV）分割中展示了有效性。

英文摘要：

Background and Purpose: Functional assessment of the left ventricle using gated myocardial perfusion (MPS) single-photon emission computed tomography relies on the precise extraction of the left ventricular contours while simultaneously ensuring the security of patient data. Methods: In this paper, we introduce the integration of Federated Domain Adaptation with TimeSformer, named 'FedDA-TSformer' for left ventricle segmentation using MPS. FedDA-TSformer captures spatial and temporal features in gated MPS images, leveraging spatial attention, temporal attention, and federated learning for improved domain adaptation while ensuring patient data security. In detail, we employed Divide-Space-Time-Attention mechanism to extract spatio-temporal correlations from the multi-centered MPS datasets, ensuring that predictions are spatio-temporally consistent. To achieve domain adaptation, we align the model output on MPS from three different centers using local maximum mean discrepancy (LMMD) loss. This approach effectively addresses the dual requirements of federated learning and domain adaptation, enhancing the model's performance during training with multi-site datasets while ensuring the protection of data from different hospitals. Results: Our FedDA-TSformer was trained and evaluated using MPS datasets collected from three hospitals, comprising a total of 150 subjects. Each subject's cardiac cycle was divided into eight gates. The model achieved Dice Similarity Coefficients (DSC) of 0.842 and 0.907 for left ventricular (LV) endocardium and epicardium segmentation, respectively. Conclusion: Our proposed FedDA-TSformer model addresses the challenge of multi-center generalization, ensures patient data privacy protection, and demonstrates effectiveness in left ventricular (LV) segmentation.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。