Segmentation of prostate and prostate zones using deep learning : A multi-MRI vendor analysis.
Segmentation of prostate and prostate zones using deep learning : A multi-MRI vendor analysis.
复制标题
DOI：
10.1007/s00066-020-01607-x
复制
发表时间：
2020-10
期刊：
影响因子：
--
通讯作者：
Stoyanova R
中科院分区：
文献类型：
--
作者：
Zavala-Romero O;Breto AL;Xu IR;Chang YC;Gautney N;Dal Pra A;Abramowitz MC;Pollack A;Stoyanova R
开发一种基于深度学习的前列腺及其周围区域 (PZ) 分割算法，该算法在多个 MRI 供应商中都是可靠的。这是一项回顾性研究。该数据集由 550 个 MRI（Siemens-330、General Electric[GE]-220）组成。多流 3D 卷积神经网络用于使用 T2 加权 (T2-w) MRI 自动分割前列腺及其 PZ。在轴向 T2-w 上手动绘制前列腺和 PZ 的轮廓。该网络使用轴向、冠状和矢状 T2-w 系列作为输入。输入数据的预处理包括偏差校正、重采样和图像归一化。来自两家 MRI 供应商（西门子和 GE）的数据集用于测试所提出的网络。训练了六种不同的模型，三种用于前列腺，三种用于前列腺区。在这三个模型中，有两个模型分别根据来自每个供应商的数据进行了训练，第三个模型（组合）则根据数据集的聚合进行了训练。 Dice 系数 (DSC) 用于比较手动分割和预测分割。对于前列腺分割，组合模型在西门子和 GE 上分别获得了 0.893 ± 0.036 和 0.825 ± 0.112（平均值 ± 标准差）的 DSC。对于 PZ，最佳 DSC 来自组合模型：0.811 ± 0.079 和 0.788 ± 0.093。虽然西门子模型在 GE 数据集上表现不佳，反之亦然，但组合模型在两个数据集上都实现了稳健的性能。所提出的网络的性能可与分割前列腺及其 PZ 的专家间变异性相媲美。结合不同 MRI 供应商的图像进行网络训练对于构建前列腺和 PZ 分割的通用模型至关重要。
Develop a deep-learning-based segmentation algorithm for prostate and its peripheral zone (PZ) that is reliable across multiple MRI vendors. This is a retrospective study. The dataset consisted of 550 MRIs (Siemens-330, General Electric[GE]-220). A multistream 3D convolutional neural network is used for automatic segmentation of the prostate and its PZ using T2-weighted (T2-w) MRI. Prostate and PZ were manually contoured on axial T2-w. The network uses axial, coronal, and sagittal T2-w series as input. The preprocessing of the input data includes bias correction, resampling, and image normalization. A dataset from two MRI vendors (Siemens and GE) is used to test the proposed network. Six different models were trained, three for the prostate and three for the PZ. Of the three, two were trained on data from each vendor separately, and a third (Combined) on the aggregate of the datasets. The Dice coefficient (DSC) is used to compare the manual and predicted segmentation. For prostate segmentation, the Combined model obtained DSCs of 0.893 ± 0.036 and 0.825 ± 0.112 (mean ± standard deviation) on Siemens and GE, respectively. For PZ, the best DSCs were from the Combined model: 0.811 ± 0.079 and 0.788 ± 0.093. While the Siemens model underperformed on the GE dataset and vice versa, the Combined model achieved robust performance on both datasets. The proposed network has a performance comparable to the interexpert variability for segmenting the prostate and its PZ. Combining images from different MRI vendors on the training of the network is of paramount importance for building a universal model for prostate and PZ segmentation.
登录
查看更多内容
DOI：
10.1007/s11548-018-1841-4
发表时间：
2018-11
影响因子：
3
作者：
To MNN;Vu DQ;Turkbey B;Choyke PL;Kwak JT
通讯作者：
Kwak JT
影响因子：
3.8
作者：
Korsager, Anne Sofie;Fortunati, Valerio;van Walsum, Theo
通讯作者：
van Walsum, Theo
影响因子：
10.6
作者：
Guo Y;Gao Y;Shen D
通讯作者：
Shen D
影响因子：
10.9
作者：
Litjens, Geert;Toth, Robert;van de Ven, Wendy;Hoeks, Caroline;Kerkstra, Sjoerd;van Ginneken, Bram;Vincent, Graham;Guillard, Gwenael;Birbeck, Neil;Zhang, Jindang;Strand, Robin;Malmberg, Filip;Ou, Yangming;Davatzikos, Christos;Kirschner, Matthias;Jung, Florian;Yuan, Jing;Qiu, Wu;Gao, Qinquan;Edwards, Philip Eddie;Maan, Bianca;van der Heijden, Ferdinand;Ghose, Soumya;Mitra, Jhimli;Dowling, Jason;Barratt, Dean;Huisman, Henkjan;Madabhushi, Anant
通讯作者：
Madabhushi, Anant
DOI：
10.1109/embc.2014.6944225
发表时间：
2014
期刊：
Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference
影响因子：
--
作者：
Cheng R;Turkbey B;Gandler W;Agarwal HK;Shah VP;Bokinsky A;McCreedy E;Wang S;Sankineni S;Bernardo M;Pohida T;Choyke P;McAuliffe MJ
通讯作者：
McAuliffe MJ