基于 U-Nets 的 T2 加权（T2W）和表观扩散系数（ADC）图磁共振成像上前列腺分区解剖结构的自动分割。

Automated segmentation of prostate zonal anatomy on T2-weighted (T2W) and apparent diffusion coefficient (ADC) map MR images using U-Nets.

机构信息

Department of Systems and Computer Engineering, Carleton University, Ottawa, ON, Canada.

Department of Radiology, University of Ottawa, Ottawa, ON, Canada.

出版信息

Med Phys. 2019 Jul;46(7):3078-3090. doi: 10.1002/mp.13550. Epub 2019 May 11.

DOI:10.1002/mp.13550

PMID:31002381

Abstract

PURPOSE

Accurate regional segmentation of the prostate boundaries on magnetic resonance (MR) images is a fundamental requirement before automated prostate cancer diagnosis can be achieved. In this paper, we describe a novel methodology to segment prostate whole gland (WG), central gland (CG), and peripheral zone (PZ), where PZ + CG = WG, from T2W and apparent diffusion coefficient (ADC) map prostate MR images.

METHODS

We designed two similar models each made up of two U-Nets to delineate the WG, CG, and PZ from T2W and ADC map MR images, separately. The U-Net, which is a modified version of a fully convolutional neural network, includes contracting and expanding paths with convolutional, pooling, and upsampling layers. Pooling and upsampling layers help to capture and localize image features with a high spatial consistency. We used a dataset consisting of 225 patients (combining 153 and 72 patients with and without clinically significant prostate cancer) imaged with multiparametric MRI at 3 Tesla.

RESULTS AND CONCLUSION

Our proposed model for prostate zonal segmentation from T2W was trained and tested using 1154 and 1587 slices of 100 and 125 patients, respectively. Median of Dice similarity coefficient (DSC) on test dataset for prostate WG, CG, and PZ were 95.33 ± 7.77%, 93.75 ± 8.91%, and 86.78 ± 3.72%, respectively. Designed model for regional prostate delineation from ADC map images was trained and validated using 812 and 917 slices from 100 and 125 patients. This model yielded a median DSC of 92.09 ± 8.89%, 89.89 ± 10.69%, and 86.1 ± 9.56% for prostate WG, CG, and PZ on test samples, respectively. Further investigation indicated that the proposed algorithm reported high DSC for prostate WG segmentation from both T2W and ADC map MR images irrespective of WG size. In addition, segmentation accuracy in terms of DSC does not significantly vary among patients with or without significant tumors.

SIGNIFICANCE

We describe a method for automated prostate zonal segmentation using T2W and ADC map MR images independent of prostate size and the presence or absence of tumor. Our results are important in terms of clinical perspective as fully automated methods for ADC map images, which are considered as one of the most important sequences for prostate cancer detection in the PZ and CG, have not been reported previously.

摘要

目的

在实现自动化前列腺癌诊断之前，准确地对磁共振（MR）图像上的前列腺边界进行区域分割是一项基本要求。本文描述了一种新的方法，用于从 T2W 和表观扩散系数（ADC）图前列腺 MR 图像中分割前列腺全腺（WG）、中央腺（CG）和外周带（PZ），其中 PZ+CG=WG。

方法

我们设计了两个类似的模型，每个模型都由两个 U-Net 组成，分别用于从 T2W 和 ADC 图 MR 图像中描绘 WG、CG 和 PZ。U-Net 是一种经过修改的全卷积神经网络，包括收缩和扩展路径，其中包含卷积、池化和上采样层。池化和上采样层有助于捕获和本地化具有高空间一致性的图像特征。我们使用了一个由 225 名患者组成的数据集（结合了 153 名和 72 名有和没有临床显著前列腺癌的患者），这些患者在 3T 进行了多参数 MRI 成像。

结果和结论

我们提出的用于从 T2W 进行前列腺分区分割的模型分别使用了 100 名和 125 名患者的 1154 和 1587 个切片进行了训练和测试。在测试数据集上，前列腺 WG、CG 和 PZ 的平均 Dice 相似系数（DSC）分别为 95.33±7.77%、93.75±8.91%和 86.78±3.72%。用于从 ADC 图图像进行区域前列腺描绘的设计模型使用了 100 名和 125 名患者的 812 和 917 个切片进行了训练和验证。该模型在测试样本上分别获得了前列腺 WG、CG 和 PZ 的平均 DSC 为 92.09±8.89%、89.89±10.69%和 86.1±9.56%。进一步的研究表明，无论 WG 大小如何，该算法在 T2W 和 ADC 图 MR 图像上报告的前列腺 WG 分割的高 DSC。此外，在 DSC 方面的分割准确性在有或没有显著肿瘤的患者之间没有显著差异。