Department of Radiological Sciences, John Tu and Thomas Yuen Center for Functional Onco-Imaging, University of California, 164 Irvine Hall, Irvine, CA, 92697-5020.
Department of Radiological Sciences, John Tu and Thomas Yuen Center for Functional Onco-Imaging, University of California, 164 Irvine Hall, Irvine, CA, 92697-5020; Department of Radiology, E-Da Hospital and I-Shou University, No. 1, Yida Road, Jiaosu Village, Yanchao District, Kaohsiung, Taiwan, 8244.
Acad Radiol. 2019 Nov;26(11):1526-1535. doi: 10.1016/j.acra.2019.01.012. Epub 2019 Jan 31.
Breast segmentation using the U-net architecture was implemented and tested in independent validation datasets to quantify fibroglandular tissue volume in breast MRI.
Two datasets were used. The training set was MRI of 286 patients with unilateral breast cancer. The segmentation was done on the contralateral normal breasts. The ground truth for the breast and fibroglandular tissue (FGT) was obtained by using a template-based segmentation method. The U-net deep learning algorithm was implemented to analyze the training set, and the final model was obtained using 10-fold cross-validation. The independent validation set was MRI of 28 normal volunteers acquired using four different MR scanners. Dice Similarity Coefficient (DSC), voxel-based accuracy, and Pearson's correlation were used to evaluate the performance.
For the 10-fold cross-validation in the initial training set of 286 patients, the DSC range was 0.83-0.98 (mean 0.95 ± 0.02) for breast and 0.73-0.97 (mean 0.91 ± 0.03) for FGT; and the accuracy range was 0.92-0.99 (mean 0.98 ± 0.01) for breast and 0.87-0.99 (mean 0.97 ± 0.01) for FGT. For the entire 224 testing breasts of the 28 normal volunteers in the validation datasets, the mean DSC was 0.86 ± 0.05 for breast, 0.83 ± 0.06 for FGT; and the mean accuracy was 0.94 ± 0.03 for breast and 0.93 ± 0.04 for FGT. The testing results for MRI acquired using four different scanners were comparable.
Deep learning based on the U-net algorithm can achieve accurate segmentation results for the breast and FGT on MRI. It may provide a reliable and efficient method to process large number of MR images for quantitative analysis of breast density.
使用 U 形网络架构进行乳房分割，并在独立验证数据集进行测试，以量化乳房 MRI 中的纤维腺体组织体积。
使用了两个数据集。训练集为 286 例单侧乳腺癌患者的 MRI。对侧正常乳房进行分割。使用基于模板的分割方法获得乳房和纤维腺体组织（FGT）的真实数据。使用 10 倍交叉验证来实现 U 形网络深度学习算法，分析训练集，并最终获得模型。独立验证集为 28 名正常志愿者的 MRI，使用四种不同的磁共振扫描仪采集。使用 Dice 相似系数（DSC）、体素精度和 Pearson 相关系数来评估性能。
在 286 例患者初始训练集的 10 倍交叉验证中，DSC 范围为 0.83-0.98（均值 0.95 ± 0.02），用于乳房；0.73-0.97（均值 0.91 ± 0.03），用于 FGT；准确性范围为 0.92-0.99（均值 0.98 ± 0.01），用于乳房；0.87-0.99（均值 0.97 ± 0.01），用于 FGT。在验证数据集的 28 名正常志愿者的 224 个测试乳房中，平均 DSC 为 0.86 ± 0.05，用于乳房；0.83 ± 0.06，用于 FGT；平均准确性为 0.94 ± 0.03，用于乳房；0.93 ± 0.04，用于 FGT。使用四种不同扫描仪获得的测试结果相当。
基于 U 形网络算法的深度学习可以实现 MRI 中乳房和 FGT 的准确分割结果。它可能为处理大量磁共振图像提供一种可靠且高效的方法，用于对乳房密度进行定量分析。
