Automatic prostate and prostate zones segmentation of magnetic resonance images using DenseNet-like U-net
Automatic prostate and prostate zones segmentation of magnetic resonance images using DenseNet-like U-net
复制标题
DOI：
10.1038/s41598-020-71080-0
复制
发表时间：
2020-08-31
影响因子：
4.6
通讯作者：
Dewey, Marc
中科院分区：
文献类型：
--
作者：
Aldoj, Nader;Biavati, Federico;Dewey, Marc
磁共振成像（MRI）提供了前列腺及其区域的详细解剖图像。它在许多诊断应用中起着至关重要的作用。自动分割，如从MR图像的前列腺和前列腺区的自动分割，促进了许多诊断和治疗应用。然而，缺乏明确的前列腺边界，前列腺组织的异质性，以及前列腺形状的广泛个体差异使这成为一个非常具有挑战性的任务。为了解决这个问题，我们提出了一个新的神经网络来自动分割前列腺及其区域。我们称这种算法为Dense U-net，因为它受到了现有的两个最先进的工具-DenseNet和U-net的启发。我们在141个患者数据集上训练了该算法，并在47个患者数据集上使用轴向T2加权图像以四重交叉验证的方式对其进行了测试。这些网络分别在弱注释和精确注释的掩码上进行训练和测试，以测试网络即使在标签不准确的情况下也可以学习的假设。该网络成功地检测到前列腺区域，并分割腺体及其区域。与U-net相比，我们算法的第二个版本Dense-2 U-net实现了整个前列腺的平均Dice评分为92.1 +/- 0.8% vs. 90.7 +/-2%，中心区为89.5 +/- 2% vs. 89.1 +/-2.2%，外周区为78.1 +/- 2.5% vs. 75 +/-3%。我们的初步结果表明，Dense-2 U-net在自动分割前列腺和前列腺区域方面比最先进的U-net更准确。
Magnetic resonance imaging (MRI) provides detailed anatomical images of the prostate and its zones. It has a crucial role for many diagnostic applications. Automatic segmentation such as that of the prostate and prostate zones from MR images facilitates many diagnostic and therapeutic applications. However, the lack of a clear prostate boundary, prostate tissue heterogeneity, and the wide interindividual variety of prostate shapes make this a very challenging task. To address this problem, we propose a new neural network to automatically segment the prostate and its zones. We term this algorithm Dense U-net as it is inspired by the two existing state-of-the-art tools-DenseNet and U-net. We trained the algorithm on 141 patient datasets and tested it on 47 patient datasets using axial T2-weighted images in a four-fold cross-validation fashion. The networks were trained and tested on weakly and accurately annotated masks separately to test the hypothesis that the network can learn even when the labels are not accurate. The network successfully detects the prostate region and segments the gland and its zones. Compared with U-net, the second version of our algorithm, Dense-2 U-net, achieved an average Dice score for the whole prostate of 92.1 +/- 0.8% vs. 90.7 +/- 2%, for the central zone of 89.5 +/- 2% vs. 89.1 +/- 2.2 %, and for the peripheral zone of 78.1 +/- 2.5% vs. 75 +/- 3%. Our initial results show Dense-2 U-net to be more accurate than state-of-the-art U-net for automatic segmentation of the prostate and prostate zones.