CAS Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China; Department of the Artificial Intelligence Technology, University of Chinese Academy of Sciences, Beijing, 100049, China.
CAS Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China; Department of the Artificial Intelligence Technology, University of Chinese Academy of Sciences, Beijing, 100049, China; Zhuhai Precision Medical Center, Zhuhai People's Hospital (affiliated with Jinan University), Zhuhai, 519000, China.
Med Image Anal. 2021 Jan;67:101873. doi: 10.1016/j.media.2020.101873. Epub 2020 Oct 18.
Automatic semantic segmentation in 2D echocardiography is vital in clinical practice for assessing various cardiac functions and improving the diagnosis of cardiac diseases. However, two distinct problems have persisted in automatic segmentation in 2D echocardiography, namely the lack of an effective feature enhancement approach for contextual feature capture and lack of label coherence in category prediction for individual pixels. Therefore, in this study, we propose a deep learning model, called deep pyramid local attention neural network (PLANet), to improve the segmentation performance of automatic methods in 2D echocardiography. Specifically, we propose a pyramid local attention module to enhance features by capturing supporting information within compact and sparse neighboring contexts. We also propose a label coherence learning mechanism to promote prediction consistency for pixels and their neighbors by guiding the learning with explicit supervision signals. The proposed PLANet was extensively evaluated on the dataset of cardiac acquisitions for multi-structure ultrasound segmentation (CAMUS) and sub-EchoNet-Dynamic, which are two large-scale and public 2D echocardiography datasets. The experimental results show that PLANet performs better than traditional and deep learning-based segmentation methods on geometrical and clinical metrics. Moreover, PLANet can complete the segmentation of heart structures in 2D echocardiography in real time, indicating a potential to assist cardiologists accurately and efficiently.
二维超声心动图中的自动语义分割在临床实践中至关重要，可用于评估各种心脏功能并提高心脏疾病的诊断水平。然而，二维超声心动图中的自动分割仍然存在两个明显的问题，即缺乏有效的上下文特征捕获的特征增强方法和类别预测中个体像素标签的不连贯性。因此，在本研究中，我们提出了一种名为深度金字塔局部注意神经网络（PLANet）的深度学习模型，以提高二维超声心动图自动方法的分割性能。具体来说，我们提出了一种金字塔局部注意模块，通过捕获紧凑和稀疏的邻域内的支持信息来增强特征。我们还提出了一种标签一致性学习机制，通过使用显式监督信号引导学习，促进像素及其邻居的预测一致性。我们在两个大规模的公共二维超声心动图数据集——心脏多结构超声分割数据集（CAMUS）和子 EchoNet-Dynamic 上对所提出的 PLANet 进行了广泛评估。实验结果表明，PLANet 在几何和临床指标上的性能优于传统和基于深度学习的分割方法。此外，PLANet 可以实时完成二维超声心动图中心脏结构的分割，表明其具有协助心脏病专家进行准确高效诊断的潜力。
