Automatic magnetic resonance prostate segmentation by deep learning with holistically nested networks

被引：59

作者：

Cheng R. ^{[1
]}

Roth H.R. ^{[2
]}

Lay N. ^{[2
]}

Lu L. ^{[2
]}

Turkbey B. ^{[3
]}

Gandler W. ^{[1
]}

McCreedy E.S. ^{[1
]}

Pohida T. ^{[4
]}

Pinto P.A. ^{[5
]}

Choyke P. ^{[3
]}

McAuliffe M.J. ^{[1
]}

Summers R.M. ^{[2
]}

机构：

[1] Imaging Sciences Laboratory, Center of Information Technology, NIH, Bethesda, MD

[2] Imaging Biomarkers and CAD Laboratory, Clinical Center, NIH, Bethesda, MD

[3] Molecular Imaging Program, NCI, Bethesda, MD

[4] Computational Bioscience and Engineering Laboratory, Center of Information Technology, NIH, Bethesda, MD

[5] Center of Cancer Research, Urologic Oncology Branch, Bethesda, MD

来源：

Journal of Medical Imaging | 2017年 / 4卷 / 04期

基金：

美国国家卫生研究院;

关键词：

deep learning; holistically nested edge detection; holistically nested networks; magnetic resonance images; prostate; segmentation;

D O I：

10.1117/1.JMI.4.4.041302

中图分类号：

学科分类号：

摘要：

Accurate automatic segmentation of the prostate in magnetic resonance images (MRI) is a challenging task due to the high variability of prostate anatomic structure. Artifacts such as noise and similar signal intensity of tissues around the prostate boundary inhibit traditional segmentation methods from achieving high accuracy. We investigate both patch-based and holistic (image-to-image) deep-learning methods for segmentation of the prostate. First, we introduce a patch-based convolutional network that aims to refine the prostate contour which provides an initialization. Second, we propose a method for end-to-end prostate segmentation by integrating holistically nested edge detection with fully convolutional networks. Holistically nested networks (HNN) automatically learn a hierarchical representation that can improve prostate boundary detection. Quantitative evaluation is performed on the MRI scans of 250 patients in fivefold cross-validation. The proposed enhanced HNN model achieves a mean ± standard deviation. A Dice similarity coefficient (DSC) of 89.77%±3.29% and a mean Jaccard similarity coefficient (IoU) of 81.59%±5.18% are used to calculate without trimming any end slices. The proposed holistic model significantly (p<0.001) outperforms a patch-based AlexNet model by 9% in DSC and 13% in IoU. Overall, the method achieves state-of-the-art performance as compared with other MRI prostate segmentation methods in the literature. © 2017 Society of Photo-Optical Instrumentation Engineers (SPIE).

引用

共 28 条

[1]

Klein S., Et al., Automatic segmentation of the prostate in 3D MR images by atlas matching using localized mutual information, Med. Phys., 35, 4, pp. 1407-1417, (2008)

[2]

Yin Y., Et al., Fully automated prostate segmentation in 3D MR based on normalized gradient fields cross-correlation initialization and LOGISMOS refinement, Proc. SPIE, 8314, (2012)

[3]

Ghose S., Et al., Texture guided active appearance model propagation for prostate segmentation, Lect. Notes Comput. Sci., 6367, pp. 111-120, (2010)

[4]

Toth R., Madabhushi A., Multifeature landmark-free active appearance models: Application to prostate MRI segmentation, IEEE Trans. Med. Imaging, 31, 8, pp. 1638-1650, (2012)

[5]

Maan B., Van der Heijden F., Futterer J.J., A new prostate segmentation approach using multispectral magnetic resonance imaging and a statistical pattern classifier, Proc. SPIE, 8314, (2012)

[6]

Habes M., Et al., Automated prostate segmentation in whole-body MRI scans for epidemiological studies, Phys. Med. Biol., 58, 17, pp. 5899-5915, (2013)

[7]

Liao S., Et al., Representation learning: A unified deep learning framework for automatic prostate MR segmentation, Int. Conf. on Medical Image Computing and Computer-Assisted Intervention (MICCAI 2013), 16, 2, pp. 254-261, (2013)

[8]

Guo Y., Gao Y., Shen D., Deformable MR prostate segmentation via deep feature learning and sparse patch matching, IEEE Trans. Med. Imaging, 35, 4, pp. 1077-1089, (2016)

[9]

Milletari F., Navab N., Ahmadi S., V-Net: Fully convolutional neural networks for volumetric medical image segmentation, IEEE, Fourth Int. Conf. on 3D Vision (3DV), pp. 565-571, (2016)

[10]

Yu L., Et al., Volumetric ConvNets with mixed residual connections for automated prostate segmentation from 3D MR images, Thirty-First AAAI Conf. on Artificial Intelligence, pp. 66-72, (2017)

← 1 2 3 →