在醫療領域當中，專業人士藉由觀察大腦皮質下不同區域體積的異變，進而做出診斷與分析，如評估病人的病況以及檢查神經系統狀況。因此，取得精準的大腦皮質下分割結果至關重要。本研究基於神經網路架構，開發具大規模資料集之模型，利用七千餘筆三維的大腦影像，搭配FreeSurfer產製之輔助標籤訓練，實現大腦皮質下之分割。其中包含四個子實驗項目，透過比較不同資料量多寡、預處理方式，篩選出表現最好的模型，且為了模擬臨床上之應用，更加入了再現性、體積精準度之評估，並與既有分割軟體FreeSurfer進行比較。根據綜合實驗的結果得出，名為「MX_RW」的模型最為出色，其在推論時期不僅不需要額外的預處理步驟，切割結果也表現出高精準度和穩定的再現性，且在體積評估方面，與手圈資料有高度的一致性。此外，MX_RW在三秒內即可切割出43個大腦皮質下區域，有利於研究人員快速地進行大腦形態學分析，並為臨床應用上提供了一套有效率的切割工具。

In the neuroimaging analysis, it is significant to increase the accuracy of segmentation techniques for subcortical brain structures. In the present study, we used the neural-network-based method to develop a large-scale model for subcortical brain segmentation by training over 7000 subject data with auxiliary labels produced using FreeSurfer segmentation. Comprehensive experiments such as varying sizes of training samples and the optimal pre-processing steps were performed to optimize the model. We further evaluated the models in the aspect of reproducibility and volumetric accuracy compared against the standard FreeSurfer pipeline. After conducting four comprehensive experiments, we have excogitated the model called "MX_RW", a superior scheme which was free from additional pre-processing steps during the testing stage, yielded high segmentation accuracy in Dice score, revealed robust reproducibility, and exhibited highly consistent with manual segmentation in volume estimates. Also, MX_RW efficiently segments subcortical regions in less than 3 seconds on GPU for a single brain volume considering the inference time. In conclusion, the proposed scheme provides researchers an efficient and accurate tool for brain morphological analysis.

1. Cahn, W., et al., Brain volume changes in first-episode schizophrenia: a 1-year follow-up study. Archives of general psychiatry, 2002. 59(11): p. 1002-1010.
2. Geevarghese, R., et al., Subcortical structure volumes and correlation to clinical variables in Parkinson's disease. Journal of Neuroimaging, 2015. 25(2): p. 275-280.
3. Renteria, M.E., et al., Subcortical brain structure and suicidal behaviour in major depressive disorder: a meta-analysis from the ENIGMA-MDD working group. Translational psychiatry, 2017. 7(5): p. e1116-e1116.
4. Holland, D., et al., Subregional neuroanatomical change as a biomarker for Alzheimer's disease. Proceedings of the National Academy of Sciences, 2009. 106(49): p. 20954-20959.
5. Goldman, S., et al., Motor stereotypies and volumetric brain alterations in children with Autistic Disorder. Research in autism spectrum disorders, 2013. 7(1): p. 82-92.
6. Scanlon, C., et al., Cortical thinning and caudate abnormalities in first episode psychosis and their association with clinical outcome. Schizophrenia research, 2014. 159(1): p. 36-42.
7. Roh, J.H., et al., Volume reduction in subcortical regions according to severity of Alzheimer’s disease. Journal of neurology, 2011. 258(6): p. 1013-1020.
8. Sánchez-Benavides, G., et al., Manual validation of FreeSurfer's automated hippocampal segmentation in normal aging, mild cognitive impairment, and Alzheimer Disease subjects. Psychiatry Research: Neuroimaging, 2010. 181(3): p. 219-225.
9. Morey, R.A., et al., A comparison of automated segmentation and manual tracing for quantifying hippocampal and amygdala volumes. Neuroimage, 2009. 45(3): p. 855-866.
10. Jenkinson, M., et al., Fsl. Neuroimage, 2012. 62(2): p. 782-790.
11. Warfield, S.K., K.H. Zou, and W.M. Wells, Simultaneous truth and performance level estimation (STAPLE): an algorithm for the validation of image segmentation. IEEE transactions on medical imaging, 2004. 23(7): p. 903-921.
12. Aljabar, P., et al., Multi-atlas based segmentation of brain images: atlas selection and its effect on accuracy. Neuroimage, 2009. 46(3): p. 726-738.
13. Fischl, B., FreeSurfer. Neuroimage, 2012. 62(2): p. 774-781.
14. Geman, S. and D. Geman, Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images. IEEE Transactions on pattern analysis and machine intelligence, 1984(6): p. 721-741.
15. Wenger, E., et al., Comparing manual and automatic segmentation of hippocampal volumes: reliability and validity issues in younger and older brains. Human brain mapping, 2014. 35(8): p. 4236-4248.
16. Kaku, A., et al., DARTS: DenseUnet-based automatic rapid tool for brain segmentation. arXiv preprint arXiv:1911.05567, 2019.
17. Krizhevsky, A., I. Sutskever, and G.E. Hinton, Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems, 2012. 25: p. 1097-1105.
18. Roy, A.G., et al. Error corrective boosting for learning fully convolutional networks with limited data. in International Conference on Medical Image Computing and Computer-Assisted Intervention. 2017. Springer.
19. Çiçek, Ö., et al. 3D U-Net: learning dense volumetric segmentation from sparse annotation. in International conference on medical image computing and computer-assisted intervention. 2016. Springer.
20. Dai, C., et al., Transfer learning from partial annotations for whole brain segmentation, in Domain Adaptation and Representation Transfer and Medical Image Learning with Less Labels and Imperfect Data. 2019, Springer. p. 199-206.
21. Coupé, P., et al., AssemblyNet: A large ensemble of CNNs for 3D whole brain MRI segmentation. NeuroImage, 2020. 219: p. 117026.
22. Wachinger, C., M. Reuter, and T. Klein, DeepNAT: Deep convolutional neural network for segmenting neuroanatomy. NeuroImage, 2018. 170: p. 434-445.
23. Huo, Y., et al., 3D whole brain segmentation using spatially localized atlas network tiles. NeuroImage, 2019. 194: p. 105-119.
24. Di Martino, A., et al., The autism brain imaging data exchange: towards a large-scale evaluation of the intrinsic brain architecture in autism. Molecular psychiatry, 2014. 19(6): p. 659-667.
25. Marcus, D.S., et al., Open Access Series of Imaging Studies (OASIS): cross-sectional MRI data in young, middle aged, nondemented, and demented older adults. Journal of cognitive neuroscience, 2007. 19(9): p. 1498-1507.
26. Klein, A., et al. Open labels: online feedback for a public resource of manually labeled brain images. in 16th Annual Meeting for the Organization of Human Brain Mapping. 2010.
27. Kennedy, D.N., et al., CANDIShare: a resource for pediatric neuroimaging data. 2012, Springer.
28. Duchesne, S., et al., The canadian dementia imaging protocol: harmonizing national cohorts. Journal of Magnetic Resonance Imaging, 2019. 49(2): p. 456-465.
29. Isensee, F., et al. Brain tumor segmentation and radiomics survival prediction: Contribution to the brats 2017 challenge. in International MICCAI Brainlesion Workshop. 2017. Springer.
30. Kingma, D.P. and J. Ba, Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014.
31. Zhou, T., S. Ruan, and S. Canu, A review: Deep learning for medical image segmentation using multi-modality fusion. Array, 2019. 3: p. 100004.
32. Yang, Y., W. Jia, and Y. Yang, Multi-atlas segmentation and correction model with level set formulation for 3D brain MR images. Pattern Recognition, 2019. 90: p. 450-463.
33. Despotović, I., B. Goossens, and W. Philips, MRI segmentation of the human brain: challenges, methods, and applications. Computational and mathematical methods in medicine, 2015. 2015.
34. Sled, J.G., A.P. Zijdenbos, and A.C. Evans, A nonparametric method for automatic correction of intensity nonuniformity in MRI data. IEEE transactions on medical imaging, 1998. 17(1): p. 87-97.
35. Avants, B.B., N. Tustison, and G. Song, Advanced normalization tools (ANTS). Insight j, 2009. 2(365): p. 1-35.
36. Van Ginneken, B., T. Heimann, and M. Styner. 3D segmentation in the clinic: A grand challenge. in MICCAI Workshop on 3D Segmentation in the Clinic: A Grand Challenge. 2007.
37. Mann, H.B. and D.R. Whitney, On a test of whether one of two random variables is stochastically larger than the other. The annals of mathematical statistics, 1947: p. 50-60.