我們提出一種用於腦瘤分割的3D卷積神經網絡架構，稱為HarDNet-BTS。HarDNet-BTS用於3D腦部磁振造影中，並分割出腦瘤中的膠質母細胞瘤的三種區域。
模型骨幹選擇使用了HarDNet，由於FC-HarDNet已經在許多應用中證明其可行性。所以我們也從此基礎下建立一個適用於3D腦瘤分割的模型架構。
我們使用 HarDNet-BTS 參加了由 RSNA-ASNR-MICCAI組織所舉辦的2021 年腦腫瘤分割挑戰賽。是一個舉辦了10年之久的比賽，且此次參加隊伍高達1200多隊。
能夠發現大家對於此任務的重視，而我們也在兢爭激烈的狀況下取得了驗證階段的第八名。
最後也受邀至BrainLes 2021 MICCAI workshop做口頭報告，與世界各地的好手共同交流，我們為台灣唯一一隊的隊伍。
我們之後會也在GitHub上公布我們的所有做法。

We propose a 3D convolutional neural network architecture for brain tumor segmentation, called HarDNet-BTS. Using HarDNet-BTS in 3D brain MRI to segment three regions of glioblastoma in brain tumors.
The model backbone chooses to use HarDNet because FC-HarDNet has proven its feasibility in many applications. Therefore, we also build a model architecture suitable for 3D brain tumor segmentation on this basis.
We used HarDNet-BTS to participate in Brain Tumor Segmentation challenge 2021 organized by RSNA-ASNR-MICCAI. This is a competition that has been running for 10 years and this time more than 1200 teams participated. It can be found that everyone attaches great importance to this task, and we also won the eighth place in the validation phase under fierce competition.
Finally, be invited to BrainLes 2021 MICCAI workshop to give oral presentations and communicate with players from all over the world. We are the only team in Taiwan. We will also publish all our practices on GitHub in the future.

摘要 i
Abstract ii
1 Introduction 1
2 Related Work 5
3 Materials and Methods 9
3.1 HarDNet-BTS Overview 9
3.2 HarDNet 10
3.3 Architecture Selection 12
3.4 Additional Improvements 18
3.4.1 Data Pre-processing 18
3.4.2 Data Augmentation 19
3.4.3 Activation Function and Downsampling 19
3.4.4 Loss Function 20
4 Experiments 21 4.1 Dataset and Experiment Setup 21
4.1.1 Dataset 21
4.1.2 Metrics 22
4.1.3 Experiment Setup 23
4.2 Results 25
4.2.1 BraTS 2021 Training Phase Result 25
4.2.2 BraTS 2021 Validation and Testing Phase Results 26
5 Conclusion and Future Work 29
5.1 Conclusion 29
5.2 Future Work 29
References 31

[1] A. Stergiou, R. Poppe, and G. Kalliatakis, “Refining activation downsampling with softpool,” in Proceedings of the IEEE/CVF International Conference on Computer Vision,
pp. 10357–10366, 2021.
[2] H.-Y. Wu and Y.-L. Lin, “Hardnet-bts: A harmonic shortcut network for brain tumor
segmentation,” in International MICCAI Brainlesion Workshop, pp. 261–271, Springer,
2022.
[3] S. Shibui, “The present status and trend of brain tumors based on the data of the brain
tumor registry of japan,” Brain and Nerve= Shinkei Kenkyu no Shinpo, vol. 64, no. 3,
pp. 286–290, 2012.
[4] O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in International Conference on Medical image computing and
computer-assisted intervention, pp. 234–241, Springer, 2015.
[5] Ö. Çiçek, A. Abdulkadir, S. S. Lienkamp, T. Brox, and O. Ronneberger, “3d u-net: learning dense volumetric segmentation from sparse annotation,” in International conference
on medical image computing and computer-assisted intervention, pp. 424–432, Springer,
2016.
[6] N. B. Le Duy Huynh, “A u-net++ with pre-trained efficientnet backbone for segmentation
of diseases and artifacts in endoscopy images and videos,” in CEUR Workshop Proceedings, vol. 2595, pp. 13–17, 2020.
[7] U. Baid, S. Ghodasara, S. Mohan, M. Bilello, E. Calabrese, E. Colak, K. Farahani,
J. Kalpathy-Cramer, F. C. Kitamura, S. Pati, et al., “The rsna-asnr-miccai brats 2021
benchmark on brain tumor segmentation and radiogenomic classification,” arXiv preprint
arXiv:2107.02314, 2021.
[8] S. Bakas, H. Akbari, A. Sotiras, M. Bilello, M. Rozycki, J. S. Kirby, J. B. Freymann,
K. Farahani, and C. Davatzikos, “Advancing the cancer genome atlas glioma mri collections with expert segmentation labels and radiomic features,” Scientific data, vol. 4, no. 1,
pp. 1–13, 2017.
[9] B. H. Menze, A. Jakab, S. Bauer, J. Kalpathy-Cramer, K. Farahani, J. Kirby, Y. Burren,
N. Porz, J. Slotboom, R. Wiest, et al., “The multimodal brain tumor image segmentation
benchmark (brats),” IEEE transactions on medical imaging, vol. 34, no. 10, pp. 1993–
2024, 2014.
[10] F. Isensee, P. F. Jäger, P. M. Full, P. Vollmuth, and K. H. Maier-Hein, “nnu-net for
brain tumor segmentation,” in International MICCAI Brainlesion Workshop, pp. 118–132,
Springer, 2020.
[11] H. Jia, C. Bai, W. Cai, H. Huang, and Y. Xia, “Hnf-netv2 for brain tumor segmentation
using multi-modal mr imaging,” arXiv preprint arXiv:2202.05268, 2022.
[12] Y. Wang, Y. Zhang, F. Hou, Y. Liu, J. Tian, C. Zhong, Y. Zhang, and Z. He, “Modalitypairing learning for brain tumor segmentation,” in International MICCAI Brainlesion
Workshop, pp. 230–240, Springer, 2020.
[13] Y. Yuan, “Automatic head and neck tumor segmentation in pet/ct with scale attention
network,” in 3D Head and Neck Tumor Segmentation in PET/CT Challenge, pp. 44–52,
Springer, 2020.
[14] J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 3431–3440, 2015.
[15] K. Kamnitsas, C. Ledig, V. F. Newcombe, J. P. Simpson, A. D. Kane, D. K. Menon,
D. Rueckert, and B. Glocker, “Efficient multi-scale 3d cnn with fully connected crf for
accurate brain lesion segmentation,” Medical image analysis, vol. 36, pp. 61–78, 2017.
[16] A. Myronenko, “3d mri brain tumor segmentation using autoencoder regularization,” in
International MICCAI Brainlesion Workshop, pp. 311–320, Springer, 2018.
[17] Z. Jiang, C. Ding, M. Liu, and D. Tao, “Two-stage cascaded u-net: 1st place solution to
brats challenge 2019 segmentation task,” in International MICCAI brainlesion workshop,
pp. 231–241, Springer, 2019.
[18] F. Isensee, P. F. Jäger, P. M. Full, P. Vollmuth, and K. H. Maier-Hein, “nnu-net for
brain tumor segmentation,” in International MICCAI Brainlesion Workshop, pp. 118–132,
Springer, 2020.
[19] H. M. Luu and S.-H. Park, “Extending nn-unet for brain tumor segmentation,” in International MICCAI Brainlesion Workshop, pp. 173–186, Springer, 2022.
[20] P. Chao, C.-Y. Kao, Y.-S. Ruan, C.-H. Huang, and Y.-L. Lin, “Hardnet: A low memory
traffic network,” in Proceedings of the IEEE/CVF international conference on computer
vision, pp. 3552–3561, 2019.
[21] G. Wang, W. Li, S. Ourselin, and T. Vercauteren, “Automatic brain tumor segmentation using cascaded anisotropic convolutional neural networks,” in International MICCAI
brainlesion workshop, pp. 178–190, Springer, 2017.
[22] F. Isensee, P. Kickingereder, W. Wick, M. Bendszus, and K. H. Maier-Hein, “Brain tumor segmentation and radiomics survival prediction: Contribution to the brats 2017 challenge,” in International MICCAI Brainlesion Workshop, pp. 287–297, Springer, 2017.
[23] F. Isensee, P. Kickingereder, W. Wick, M. Bendszus, and K. H. Maier-Hein, “No new-net,”
in International MICCAI Brainlesion Workshop, pp. 234–244, Springer, 2018.
[24] C. Zhou, S. Chen, C. Ding, and D. Tao, “Learning contextual and attentive information for
brain tumor segmentation,” in International MICCAI brainlesion workshop, pp. 497–507,
Springer, 2018.
[26] R. McKinley, M. Rebsamen, R. Meier, and R. Wiest, “Triplanar ensemble of 3d-to-2d cnns
with label-uncertainty for brain tumor segmentation,” in International MICCAI brainlesion workshop, pp. 379–387, Springer, 2019.
[27] H. Jia, W. Cai, H. Huang, and Y. Xia, “H2nf-net for brain tumor segmentation using multimodal mr imaging: 2nd place solution to brats challenge 2020 segmentation task,” in
Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries-6th International Workshop, Springer, pp. 58–68, 2021.
[28] Y. Yuan, “Evaluating scale attention network for automatic brain tumor segmentation with
large multi-parametric mri database,” in International MICCAI Brainlesion Workshop,
pp. 42–53, Springer, 2022.
[29] M. Futrega, A. Milesi, M. Marcinkiewicz, and P. Ribalta, “Optimized u-net for brain tumor segmentation,” in International MICCAI Brainlesion Workshop, pp. 15–29, Springer,
2022.
[30] D. Misra, “Mish: A self regularized non-monotonic neural activation function,” arXiv
preprint arXiv:1908.08681, vol. 4, no. 2, pp. 10–48550, 2019.
[31] C.-Y. Lee, S. Xie, P. Gallagher, Z. Zhang, and Z. Tu, “Deeply-supervised nets,” in Artificial intelligence and statistics, pp. 562–570, PMLR, 2015.
[32] G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, “Densely connected convolutional networks,” in Proceedings of the IEEE conference on computer vision and pattern
recognition, pp. 4700–4708, 2017.
[33] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in
Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770–
778, 2016.
[34] C.-H. Huang, H.-Y. Wu, and Y.-L. Lin, “Hardnet-mseg: A simple encoder-decoder polyp
segmentation neural network that achieves over 0.9 mean dice and 86 fps,” arXiv preprint
arXiv:2101.07172, 2021.
[35] V. Nair and G. E. Hinton, “Rectified linear units improve restricted boltzmann machines,”
in Icml, 2010.
[36] P. Micikevicius, S. Narang, J. Alben, G. Diamos, E. Elsen, D. Garcia, B. Ginsburg,
M. Houston, O. Kuchaiev, G. Venkatesh, et al., “Mixed precision training,” arXiv preprint
arXiv:1710.03740, 2017.
[37] T. Henry, A. Carré, M. Lerousseau, T. Estienne, C. Robert, N. Paragios, and E. Deutsch,
“Brain tumor segmentation with self-ensembled, deeply-supervised 3d u-net neural networks: a brats 2020 challenge solution,” in International MICCAI Brainlesion Workshop,
pp. 327–339, Springer, 2020.
[38] T.-Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollár, “Focal loss for dense object detection,” in Proceedings of the IEEE international conference on computer vision, pp. 2980–
2988, 2017.
[39] F. Milletari, N. Navab, and S.-A. Ahmadi, “V-net: Fully convolutional neural networks
for volumetric medical image segmentation,” in 2016 fourth international conference on
3D vision (3DV), pp. 565–571, IEEE, 2016.
[40] M. Drozdzal, E. Vorontsov, G. Chartrand, S. Kadoury, and C. Pal, “The importance of
skip connections in biomedical image segmentation,” in Deep learning and data labeling
for medical applications, pp. 179–187, Springer, 2016.
[41] L. Wright, “Ranger - a synergistic optimizer..” https://github.com/lessw2020/
Ranger-Deep-Learning-Optimizer, 2019.