Jinghui Chu, Xiaochuan Li, Jiaqi Zhang, Wei Lü. Fine-Granted Segmentation Method for Three-Dimensional Brain Tumors Using Cascaded Convolutional Network[J]. Laser & Optoelectronics Progress, 2019, 56(10): 101001
- Laser & Optoelectronics Progress
- Vol. 56, Issue 10, 101001 (2019)
Fig. 1. Position relationship amongwhole tumor, tumor core and enhancing tumor
Fig. 2. Segmentation system based on cascaded 2.5D convolutional neural network
Fig. 3. Structural diagram of proposed 2.5D V-Net
Fig. 4. Structural diagram of whole tumor segmentation module
Fig. 5. Structural diagram of tumor core segmentation module
Fig. 6. Structural diagram of enhancing tumor segmentation module
Fig. 7. MRI images in three sections for different modes. (a) T1 images; (b) T2 images; (c) FLAIR images; (d) T1ce images; (e) label mask for fine segmentation
Fig. 8. Segmentation results of different networks in three directions for sample 1. (a) T1ce images; (b) end-to-end V-Net; (c) cascaded V-Net-3D; (d) cascaded V-Net-2D; (e) cascaded V-Net-2.5D
Fig. 9. Segmentation results of different networks in three directions for sample 2. (a) T1ce images; (b) end-to-end V-Net; (c) cascaded V-Net-3D;(d) cascaded V-Net-2D; (e) cascaded V-Net-2.5D
Fig. 10. Qualitative comparison of segmentation results before and after fusion. (a) FLAIR images; (b) T1ce images; (c) horizontal prediction; (d) coronal prediction; (e) sagittal prediction; (f) fused prediction
|
Table 1. Comparison of various algorithms for ten-fold cross-validation
|
Table 2. Segmentation performance comparison of 2.5D V-Net fed by images in different directions
|
Table 3. Segmentation performance comparison based on different loss functions
Set citation alerts for the article
Please enter your email address
© Copyright 2018-2021 | Chinese Laser Press. All Rights Reserved 沪ICP备15018463号-20