03_特集2_3校_0929.indd

Size: px

Start display at page:

Download "03_特集2_3校_0929.indd"

みりあわくや
5 years ago
Views:

1 MEDICAL IMAGING TECHNOLOGY Vol. 35 No. 4 September CT 1 1 convolutional neural network; ConvNet CT CT ConvNet 2D ConvNet CT ConvNet CT CT Med Imag Tech 35 4 : , CT MR zxrgoo@gmail.com

188 Med Imag Tech Vol. 35 No. 4 September 2017 1 CT 2.

2 188 Med Imag Tech Vol. 35 No. 4 September CT 2. semantic segmentation bounding box detection classification fully convolutional networks FCN FCN 2 FCN 1 FCN ILSVRC 1 AlexNet 3 VGG 4 GoogLeNet ConvNet fully connected layers convolutionalization 1 heat-map 1

3 Med Imag Tech Vol. 35 No. 4 September De-ConvNet 1 De-convolution FCN FCN ConvNet fine-tuning transfer learning FCN De-ConvNet De-convolution FCN FCN FCN Up-pooling SegNet 6 CRF-RNN FCN +CRF-RNN 7 3. FCN CT CT semantic segmentation CT MR D-FCN 8 CT 2 CT CT availabilityreliability CT CT CT FCN 2 3D-FCN 9, 10 FCN 3D U-Net 9

4 190 Med Imag Tech Vol. 35 No. 4 September D-FCN 8 FCN 3 CT CT D U-Net CT sliding window FCN CT CT 3D U-Nets cascade CT coarse-to-fine D-FCNs 11, 12 3D-FCN CT 11, 12 CT 11 3D-FCN 12 3D-FCN 12 3D-FCN CT dice index 3D-FCN Loss 12

Med Imag Tech Vol. 35 No. 4 September 2017 191 3 CT 8 1 Ground truth Segmentation results 4. CT 3 2D-FCN 3D-FCNs 3D-FCN 3D U-Net 9 13 2009 2013 2 DB DB CT 240 512 512 99 1141 voxels 0.625 1.148 0.

5 Med Imag Tech Vol. 35 No. 4 September CT 8 1 Ground truth Segmentation results 4. CT 3 2D-FCN 3D-FCNs 3D-FCN 3D U-Net DB DB CT voxels mm CT Graphics Processing Unit NVIDIA GeForce GTX Titan X 12 GB Caffe Framework D-FCN 3D U-Net 3D U-Net CT

6 192 Med Imag Tech Vol. 35 No. 4 September 2017 CT 3D U-Net 19 2D-FCN 13.2 CT 3D U-Net CT 3D-FCN 3D-FCNs 3D-FCN 2D-FCN D-FCN CT 1 2D-FCN 6. CT CT 14 FCN FCN Holger Roth JSPS C LeCun Y, Bottou L, Bengio Y, et al.: Gradient-based learning applied to document recognition. Proc. of the IEEE 86: , Long J, Shelhamer E, Darrell T: Fully convolutional networks for semantic segmentation. In proceedings of the IEEE Conference on Computer Vision and Pattern Recognition CVPR, Boston, 2015, pp Krizhevsky A, Sutskever I, Hinton GE, et al.: Image Net classification with deep convolutional neural networks. In proceedings of Advances in Neural Information Processing Systems 25, Nevada, 2012, pp Simonyan K, Zisserman A: Very deep convolutional networks for large-scale image recognition, Proc. ICLR, accessed Szegedy C, Liu W, Jia Y, et al.: Going deeper with convolutions. In proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition CVPR, Boston, 2015, No Badrinarayannan V, Kendall A, Cipolla R: SegNet: A deep convolutional encoder-decoder architecture for image segmentation. IEEE Trans Patt Anal Mach Intell, 2017 in press 7 Zheng S, Jayasumana S, Romera-Paredes B, et al.: Conditional random fields as recurrent neural networks. In proceedings of 2015 IEEE International Conference on Computer Vision ICCV, Santiago, 2015, No Zhou X, Ito T, Takayama R, et al.: First trial and evaluation of anatomical structure segmentations in 3D CT images based only on deep learning. Med Image Inf Sci 33: 69-74, Çiçek Ö, Abdulkadir A, Lienkamp SS, et al.: 3D U- Net: Learning dense volumetric segmentation from sparse annotation. In Ourselin S, Joskowicz L, Sabuncu ML, et al. eds.: Medical Image Computing and Computer-Assisted Intervention MICCAI 2016, Lecture Notes in Computer Science Vol. 9901, Springer, Cham, pp Roth HR, Oda H, Hayashi Y, et al.: Hierarchical 3D fully convolutional networks for multi-organ segmentation. (accessed ) 11 Zhou X, Morita S, Zhou X, et al.: Automatic anatomy partitioning of the torso region on CT images by using multiple organ localizations with a group-wise calibration technique. Proc SPIE 9414: 94143K K-6, D-Deep CNN CT OP Okada T, Linguraru MG, Hori M, et al.: Abdominal

Med Imag Tech Vol. 35 No. 4 September 2017 193 multi-organ segmentation from CT images using conditional shape-location and unsupervised intensity priors.

Med Phys, 2017 in press Simultaneous Recognition and Segmentation of Multiple Anatomical Structures on CT Images by Using Deep Learning Approach Xiangrong ZHOU 1, Hiroshi FUJITA 1 1 Gifu University

7 Med Imag Tech Vol. 35 No. 4 September multi-organ segmentation from CT images using conditional shape-location and unsupervised intensity priors. Med Image Anal 26: 1-18, Zhou X, Takayama R, Wang S, et al.: Deep learning of the sectional appearances of 3D CT images for anatomical structure segmentation based on an FCN voting method. Med Phys, 2017 in press Simultaneous Recognition and Segmentation of Multiple Anatomical Structures on CT Images by Using Deep Learning Approach Xiangrong ZHOU 1, Hiroshi FUJITA 1 1 Gifu University This paper introduces research works that apply deep learning approaches based on ConvNet to solve automatic multi-organ segmentations on CT images that cover a wide range of human body. In particular, we describe our recent research work as an example to show multiple-organ segmentation methods on CT images by using ConvNets. We discuss strength and weakness of the ConvNet that is majorly used for 2D image processing and its extension for 3D images with the latest research progresses. Finally, we compare the deep learning approaches to the conventional approach that is designed by the processing procedures based on human experience and shows an advantage and potential possibility of ConvNets to address the issue of automatic multi-organ segmentations on CT images covering a wide range of human body. Key words: Deep learning, Convolutional neural network, 3D CT images, Anatomical structures recognition and extraction Med Imag Tech 35 4 : , IEEE SPIE * * *

35_3_9.dvi

35_3_9.dvi 180 Vol. 35 No. 3, pp.180 185, 2017 Image Recognition by Deep Learning Hironobu Fujiyoshi and Takayoshi Yamashita Chubu University 1. 1990 2000 Scale-Invariant Feature Transform SIFT Histogram of Oriented