Multi-target Interactive Neural Network for Automated Segmentation of the Hippocampus in Magnetic Resonance Imaging
- 23 Downloads
The hippocampus has been recognized as an important biomarker for the diagnosis and assessment of neurological diseases. Convenient and accurate automated segmentation of the hippocampus facilitates the analysis of large-scale neuroimaging studies. This work describes a novel technique for hippocampus segmentation in magnetic resonance images, in which interactive neural network (Inter-Net) is based on 3D convolutional operations. Inter-Net achieves the interaction through two aspects: one is the compartments, which builds an exponential ensemble network that integrates numerous short networks together when forward propagation. The other is the pathways, which realizes inter-connection between feature extraction and restoration. In addition, a multi-target architecture is proposed by designing multiple objective functions in terms of evaluation index, information theory, and data distribution. The proposed architecture is validated in fivefold cross-validation on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset, where the mean Dice similarity indices of 0.919 (± 0.023) and precision of 0.926 (± 0.032) for the hippocampus segmentation. The running time is approximately 42.1 s from reading the image to outputting the segmentation result in our computer configuration. We compare the experimental results of a variety of methods to prove the effectiveness of the Inter-Net and contrast integrated architectures with different objective functions to illustrate the robustness of the fusion. The proposed framework is general and can be easily extended to numerous tissue segmentation tasks while it is tailored for the hippocampus.
KeywordsHippocampus Interactive neural network Magnetic resonance images Multi-target Objective function
This work was supported by National Natural Science Foundation of China (61471064), National Science and Technology Major Project of China (No.2017ZX03001022), and BUPT Excellent Ph.D. Students Foundation (No.CX2019309).
Compliance with Ethical Standards
Conflict of Interest
The authors declare that they have no conflict of interest.
This article does not contain any studies with human participants or animals performed by any of the authors.
- 14.Chupin M, Mukuna-Bantumbakulu AR, Hasboun D, Bardinet E, Baillet S, Kinkingnéhun S, et al. Anatomically constrained region deformation for the automated segmentation of the hippocampus and the amygdala: method and validation on controls and patients with Alzheimer’s disease. Neuroimage. 2007;34(3):996–1019.CrossRefGoogle Scholar
- 26.Moghaddam MJ, Soltanian-Zadeh H. Automatic segmentation of brain structures using geometric moment invariants and artificial neural networks//International conference on Information Processing in Medical Imaging. Berlin: Springer; 2009. p. 326–37.Google Scholar
- 27.Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation. Proceedings of the IEEE conference on computer vision and pattern recognition, 2015. pp. 3431–3440.Google Scholar
- 28.Ronneberger O, Fischer P, Brox T. U-net: convolutional networks for biomedical image segmentation. International conference on medical image computing and computer-assisted intervention. Cham: Springer; 2015. p. 234–41.Google Scholar
- 30.Liu X, Deng Z. Segmentation of drivable road using deep fully convolutional residual network with pyramid pooling. Cogn Comput. 2017:1–10.Google Scholar
- 33.Yuan Y, Xun G, Ma F, et al. Muvan: a multi-view attention network for multivariate temporal data. 2018 IEEE International Conference on Data Mining (ICDM). Piscataway: IEEE; 2018. p. 717–26.Google Scholar
- 37.Chen Y, Shi B, Wang Z, Zhang P, Smith CD, Liu J. Hippocampus segmentation through multi-view ensemble ConvNets[C]//Biomedical Imaging (ISBI 2017), 2017 IEEE 14th International Symposium on. IEEE, 2017. pp. 192–196.Google Scholar
- 41.Veit A, Wilber M, Belongie S. Residual networks are exponential ensembles of relatively shallow networks. arXiv preprint. arXiv preprint arXiv:1605.06431. 2016;1(2):3.Google Scholar
- 42.He K, Zhang X, Ren S, Sun J. Identity mappings in deep residual networks. European Conference on Computer Vision. Cham: Springer; 2016. p. 630–45.Google Scholar
- 43.Nair V, Hinton G E. Rectified linear units improve restricted boltzmann machines. Proceedings of the 27th international conference on machine learning (ICML-10). 2010. pp. 807–814.Google Scholar
- 44.Zeiler MD. ADADELTA: an adaptive learning rate method. arXiv preprint arXiv:1212.5701. 2012.Google Scholar
- 45.Kingma DP, Ba J. Adam: a method for stochastic optimization. arXiv preprint arXiv:1412.6980. 2014.Google Scholar
- 46.Dauphin Y, de Vries H, Bengio Y. Equilibrated adaptive learning rates for non-convex optimization[C]. Adv Neural Inf Proces Syst. 2015:1504–12.Google Scholar
- 47.Srivastava N, Hinton G, Krizhevsky A, Stuskever I, Salakhutdinov R. Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res. 2014;15(1):1929–58.Google Scholar
- 48.Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556. 2014.Google Scholar
- 49.Zeng D, Zhao F, Shen W, Ge S. Compressing and accelerating neural network for facial point localization. Cogn Comput. 2017:1–9.Google Scholar
- 54.He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. Proceedings of the IEEE conference on computer vision and pattern recognition. 2016. pp. 770–778.Google Scholar