Department of Systems and Computer Engineering, Carleton University, Ottawa, ON, Canada.
Department of Computing and Medicine, Imperial College London, London, ON, Canada.
Med Phys. 2020 Apr;47(4):1645-1655. doi: 10.1002/mp.14022. Epub 2020 Feb 10.
Three-dimensional (3D) late gadolinium enhancement magnetic resonance (LGE-MR) imaging enables the quantification of myocardial scar at high resolution with unprecedented volumetric visualization. Automated segmentation of myocardial scar is critical for the potential clinical translation of this technique given the number of tomographic images acquired.
In this paper, we describe the development of cascaded multi-planar U-Net (CMPU-Net) to efficiently segment the boundary of the left ventricle (LV) myocardium and scar from 3D LGE-MR images. In this approach, two subnets, each containing three U-Nets, were cascaded to first segment the LV myocardium and then segment the scar within the presegmented LV myocardium. The U-Nets were trained separately using two-dimensional (2D) slices extracted from axial, sagittal, and coronal slices of 3D LGE-MR images. We used 3D LGE-MR images from 34 subjects with chronic ischemic cardiomyopathy. The U-Nets were trained using 8430 slices, extracted in three orthogonal directions from 18 images. In the testing phase, the outputs of U-Nets of each subnet were combined using the majority voting system for final label prediction of each voxel in the image. The developed method was tested for accuracy by comparing its results to manual segmentations of LV myocardium and LV scar from 7250 slices extracted from 16 3D LGE-MR images. Our method was also compared to numerous alternative methods based on machine learning, energy minimization, and intensity-thresholds.
Our algorithm reported a mean dice similarity coefficient (DSC), absolute volume difference (AVD), and Hausdorff distance (HD) of 85.14% ± 3.36%, 43.72 ± 27.18 cm , and 19.21 ± 4.74 mm for determining the boundaries of LV myocardium from LGE-MR images. Our method also yielded a mean DSC, AVD, and HD of 88.61% ± 2.54%, 9.33 ± 7.24 cm , and 17.04 ± 9.93 mm for LV scar segmentation on the unobserved test dataset. Our method significantly outperformed the alternative techniques in segmentation accuracy (P < 0.05).
The CMPU-Net method provided fully automated segmentation of LV scar from 3D LGE-MR images and outperformed the alternative techniques.
三维（3D）晚期钆增强磁共振（LGE-MR）成像能够以高分辨率实现心肌瘢痕的量化，具有前所未有的容积可视化。鉴于获取的断层图像数量众多，自动化分割心肌瘢痕对于该技术的潜在临床转化至关重要。
在本文中，我们描述了级联多平面 U-Net（CMPU-Net）的开发，用于从 3D LGE-MR 图像中有效地分割左心室（LV）心肌和瘢痕的边界。在这种方法中，两个子网，每个子网包含三个 U-Nets，首先级联分割 LV 心肌，然后在预分割的 LV 心肌内分割瘢痕。U-Nets 使用从 3D LGE-MR 图像的轴向、矢状和冠状切片中提取的二维（2D）切片分别进行训练。我们使用 34 名慢性缺血性心肌病患者的 3D LGE-MR 图像。U-Nets 使用从 18 张图像的三个正交方向提取的 8430 张切片进行训练。在测试阶段，使用多数投票系统组合每个子网的 U-Nets 的输出，以对图像中每个体素进行最终标签预测。通过将其结果与从 16 张 3D LGE-MR 图像中提取的 7250 张切片的手动 LV 心肌和 LV 瘢痕分割进行比较，测试了所开发方法的准确性。我们的方法还与基于机器学习、能量最小化和强度阈值的众多替代方法进行了比较。
我们的算法报告了从 LGE-MR 图像确定 LV 心肌边界的平均骰子相似系数（DSC）、绝对体积差异（AVD）和 Hausdorff 距离（HD）分别为 85.14%±3.36%、43.72±27.18cm和 19.21±4.74mm。我们的方法还在未观察到的测试数据集上的 LV 瘢痕分割中产生了平均 DSC、AVD 和 HD，分别为 88.61%±2.54%、9.33±7.24cm 和 17.04±9.93mm。我们的方法在分割准确性方面明显优于替代技术（P<0.05）。
CMPU-Net 方法提供了从 3D LGE-MR 图像自动分割 LV 瘢痕的方法，并优于替代技术。
