SimLVSeg:利用自监督与弱监督学习简化二维加时间超声心动图左心室分割
SimLVSeg是一种基于视频网络的新方法,通过自监督预训练和弱监督学习,从稀疏标注的超声心动图视频中实现一致的左心室分割。在EchoNet-Dynamic数据集上,其Dice分数达到93.32%(95%CI 93.21-93.43%),优于现有最先进方法;在CAM US数据集上的分布外测试显示了泛化能力。
要点
- SimLVSeg在EchoNet-Dynamic上Dice分数为93.32%,优于现有方法且计算效率更高。
- 该方法利用稀疏标注(每例仅两帧)即可实现视频级一致的左心室分割。
- 在CAM US数据集上的分布外测试展示了模型对分布偏移的泛化能力。
核心结论
- SimLVSeg在左心室分割上表现优异且计算成本较低,表明采用视频网络进行左心室分割是实现可靠分割的有前景的研究方向。
研究设计与样本
- 研究提出SimLVSeg两阶段训练范式:第一阶段采用时间掩码的自监督预训练,从大量未标注帧中捕获心动周期模式;第二阶段采用弱监督学习从稀疏标注中分割左心室。在EchoNet-Dynamic数据集上评估,并在CAM US数据集上进行分布外测试。
诊断标准与关键数值
- 在EchoNet-Dynamic数据集上,SimLVSeg的Dice分数为93.32%(95%置信区间:93.21-93.43%),优于现有最先进解决方案且更高效。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
OBJECTIVE: Achieving reliable automatic left ventricle (LV) segmentation from echocardiograms is challenging due to the inherent sparsity of annotations in the dataset, as clinicians typically only annotate two specific frames for diagnostic purposes. Here we aim to address this challenge by introducing simplified LV segmentation (SimLVSeg), a novel paradigm that enables video-based networks for consistent LV segmentation from sparsely annotated echocardiogram videos.
METHODS: SimLVSeg consists of two training stages: (i) self-supervised pre-training with temporal masking, which involves pre-training a video segmentation network by capturing the cyclic patterns of echocardiograms from largely unannotated echocardiogram frames, and (ii) weakly supervised learning tailored for LV segmentation from sparse annotations.
RESULTS: We extensively evaluated SimLVSeg using EchoNet-Dynamic, the largest echocardiography dataset. SimLVSeg outperformed state-of-the-art solutions by achieving a 93.32% (95% confidence interval: 93.21-93.43%) dice score while being more efficient. We further conducted an out-of-distribution test to showcase SimLVSeg's generalizability on distribution shifts (CAM US dataset).
CONCLUSION: Our findings show that SimLVSeg exhibits excellent performance on LV segmentation with a relatively cheaper computational cost. This suggests that adopting video-based networks for LV segmentation is a promising research direction to achieve reliable LV segmentation. Our code is publicly available at https://github.com/BioMedIA-MBZUAI/SimLVSeg.
引用
- Maani F, Ukaye A, Saadi N, et al. SimLVSeg: Simplifying Left Ventricular Segmentation in 2-D+Time Echocardiograms With Self- and Weakly Supervised Learning. Ultrasound Med Biol. 2024. PMID: 39343627. doi:10.1016/j.ultrasmedbio.2024.08.023
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。