Abstract
Quantitative susceptibility mapping (QSM) is a magnetic resonance imaging technique used to quantitatively measure the iron content in the brain. Patients with Parkinson’s disease are reported having increased iron deposition, especially in substantia nigra (SN) which is a relatively small gray matter structure located in the midbrain. The automatic segmentation of SN is a critical prerequisite step to facilitate the progression of evaluating the course of Parkinson’s disease. However, the imaging protocol and reconstruction methods in QSM acquisition vary largely, rendering great challenges in constructing and applying image segmentation models. Thus, a model trained on a certain dataset often performs poorly on datasets from other scanners or reconstruction methods. To quickly transfer a trained segmentation model to a dataset acquired in a new instrument, we have developed a joint appearance-feature domain adaptation framework (JAFDAF) to transfer the knowledge from the source to the target domains for improved SN segmentation. In particular, we perform domain adaption in both appearance and feature spaces. In the appearance space, we use region-based histogram matching and a neural network to align the grayscale ranges of images between these two domains. In the feature space, we propose a domain regularization layer (DRL) by utilizing the idea of neural architecture search (NAS) to enforce the convolution kernels for learning features that are efficacious in both domains. Ablation experiments have been carried out to evaluate the proposed JAFDAF framework, and the experimental results on 27 subjects show that our method achieves up to 12% over the baseline model and about 5% over a fine-tuning approach.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Langkammer, C., Schweser, F., Krebs, N., et al.: Quantitative susceptibility mapping (QSM) as a means to measure brain iron? A post mortem validation study. Neuroimage 62(3), 1593–1599 (2012)
Long, J., Shelhamer, E., Darrell, T.: Fully convolutional networks for semantic segmentation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 3431–3440 (2015)
Pan, S.J., Tsang, I.W., Kwok, J.T., et al.: Domain adaptation via transfer component analysis. IEEE Trans. Neural Netw. 22(2), 199–210 (2010)
Ghifary, Muhammad, Kleijn, W.B., Zhang, Mengjie: Domain adaptive neural networks for object recognition. In: Pham, Duc-Nghia, Park, Seong-Bae (eds.) PRICAI 2014. LNCS (LNAI), vol. 8862, pp. 898–904. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-13560-1_76
Tzeng, E., Hoffman, J., Zhang, N., et al.: Deep domain confusion: Maximizing for domain invariance. arXiv preprint arXiv:1412.3474 (2014)
Long, M., Cao, Y., Wang, J., et al.: Learning transferable features with deep adaptation networks. arXiv preprint arXiv:1502.02791 (2015)
Goodfellow, I., Pouget-Abadie, J., Mirza, M., et al.: Generative adversarial nets. In: Advances in Neural Information Processing Systems, pp. 2672–2680 (2014)
Li, Minjun, Huang, Haozhi, Ma, Lin, Liu, Wei, Zhang, Tong, Jiang, Yugang: Unsupervised image-to-image translation with stacked cycle-consistent adversarial networks. In: Ferrari, Vittorio, Hebert, Martial, Sminchisescu, Cristian, Weiss, Yair (eds.) ECCV 2018. LNCS, vol. 11213, pp. 186–201. Springer, Cham (2018). https://doi.org/10.1007/978-3-030-01240-3_12
Zhu, J.Y., Park, T., Isola, P., et al.: Unpaired image-to-image translation using cycle-consistent adversarial networks. In: Proceedings of the IEEE International Conference on Computer Vision, pp. 2223–2232 (2017)
Hu, J., Shen, L., Sun, G.: Squeeze-and-excitation networks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 7132–7141 (2018)
Liu, H., Simonyan, K., Yang, Y.: Darts: Differentiable architecture search. arXiv preprint arXiv:1806.09055 (2018)
Bug, Daniel, et al.: Combined learning for similar tasks with domain-switching networks. In: Shen, Dinggang, Liu, Tianming, Peters, Terry M., Staib, Lawrence H., Essert, Caroline, Zhou, Sean, Yap, Pew-Thian, Khan, Ali (eds.) MICCAI 2019. LNCS, vol. 11768, pp. 565–572. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32254-0_63
Huang, G., Liu, Z., Van Der Maaten, L., et al.: Densely connected convolutional networks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 4700–4708 (2017)
Jégou, S., Drozdzal, M., Vazquez, D., et al.: The one hundred layers tiramisu: Fully convolutional densenets for semantic segmentation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp. 11–19 (2017)
Li, W., Wu, B., Liu, C.: Quantitative susceptibility mapping of human brain reflects spatial variation in tissue composition. Neuroimage 55, 1645–1656 (2011)
Tang, J., Liu, S., Neelavalli, J., et al.: Improving susceptibility mapping using a threshold-based K-space/image domain iterative reconstruction approach. Magnet. Resonance Med. 69(5), 1396–1407 (2013)
Avants, B.B., Tustison, N., Song, G.: Advanced normalization tools (ANTS). Insight J 2(365), 1–35 (2009)
Xiao, B., He, N., Wang, Q., et al.: Quantitative susceptibility mapping based hybrid feature extraction for diagnosis of Parkinson’s disease. NeuroImage: Clin. 24, 102070 (2019)
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Xiao, B. et al. (2020). Joint Appearance-Feature Domain Adaptation: Application to QSM Segmentation Transfer. In: Liu, M., Yan, P., Lian, C., Cao, X. (eds) Machine Learning in Medical Imaging. MLMI 2020. Lecture Notes in Computer Science(), vol 12436. Springer, Cham. https://doi.org/10.1007/978-3-030-59861-7_25
Download citation
DOI: https://doi.org/10.1007/978-3-030-59861-7_25
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-59860-0
Online ISBN: 978-3-030-59861-7
eBook Packages: Computer ScienceComputer Science (R0)