Dlung: Unsupervised Few-Shot Diffeomorphic Respiratory Motion Modeling

doi:10.1007/s12204-022-2525-3

Abstract

Abstract: Lung image registration plays an important role in lung analysis applications, such as respiratory motion modeling. Unsupervised learning-based image registration methods that can compute the deformation without the requirement of supervision attract much attention. However, it is noteworthy that they have two drawbacks: they do not handle the problem of limited data and do not guarantee diffeomorphic (topologypreserving) properties, especially when large deformation exists in lung scans. In this paper, we present an unsupervised few-shot learning-based diffeomorphic lung image registration, namely Dlung. We employ fine-tuning techniques to solve the problem of limited data and apply the scaling and squaring method to accomplish the diffeomorphic registration. Furthermore, atlas-based registration on spatio-temporal (4D) images is performed and thoroughly compared with baseline methods. Dlung achieves the highest accuracy with diffeomorphic properties. It constructs accurate and fast respiratory motion models with limited data. This research extends our knowledge of respiratory motion modeling.

Key words: unsupervised few-shot learning, respiratory motion modeling, diffeomorphic registration

摘要： 肺部图像配准在肺部分析应用中具有重要作用，比如呼吸运动建模。基于无监督学习的图像配准方法可以在不需要监督的情况下计算变形，因此受到了广泛的关注。但是，需要注意的是，它们有两个缺点：它们不能处理数据不足的问题，也不能保证微分同胚（保留拓扑结构）属性，特别是当肺部扫描中存在较大形变时。本文提出了一种基于无监督少样本学习的微分同胚肺部图像配准方法，称为Dlung。我们采用微调解决数据不足的问题，并采用缩放-平方层来实现微分同胚配准。在实验中，我们进行了空间时间（4D）图像上的配准，并与基准方法进行了全面比较。Dlung取得了最高的微分同胚准确率，它使用有限的数据构建了准确和快速的呼吸运动模型。这项研究扩展了我们对呼吸运动建模的认识。

关键词: 无监督少样本学习，呼吸运动建模，微分同胚配准

CLC Number:

CHEN Peizhi1,2* (陈培芝), GUO Yifan1 (郭逸凡)，WANG Dahan1,2 (王大寒), CHEN Chinling1,3,4* (陈金铃). Dlung: Unsupervised Few-Shot Diffeomorphic Respiratory Motion Modeling[J]. J Shanghai Jiaotong Univ Sci, 2023, 28(4): 536-.

References

[1] STEWART B W, WILD C P. World cancer report 2014[M]. Lyon: International Agency for Research on Cancer, 2014.
[2] SHEN D G, DAVATZIKOS C. HAMMER: Hierarchical attribute matching mechanism for elastic registration [J]. IEEE Transactions on Medical Imaging, 2002,21(11): 1421-1439.
[3] KLEIN S, STARING M, MURPHY K, et al. Elastix:A toolbox for intensity-based medical image registration [J]. IEEE Transactions on Medical Imaging, 2010,29(1): 196-205.
[4] BEG M F, MILLER M I, TROUV′E A, et al. Computing large deformation metric mappings via geodesic flows of diffeomorphisms [J]. International Journal ofComputer Vision, 2005, 61(2): 139-157.
[5] ASHBURNER J. A fast diffeomorphic image registration algorithm [J]. NeuroImage, 2007, 38(1): 95-113.
[6] AVANTS B B, EPSTEIN C L, GROSSMAN M, etal. Symmetric diffeomorphic image registration with cross-correlation: Evaluating automated labeling of elderly and neurodegenerative brain [J]. Medical ImageAnalysis, 2008, 12(1): 26-41.
[7] JADERBERG M, SIMONYAN K, ZISSERMAN A,et al. Spatial transformer networks [M]//Advances inneural information processing systems 28. Red Hook:Curran Associates, 2015: 1-9.
[8] SOKOOTI H, VOS B D, BERENDSEN F, et al. Nonrigid image registration using multi-scale 3D convolutional neural networks [M]//Medical image computingand computer assisted intervention – MICCAI 2017.Cham: Springer, 2017: 232-239.
[9] EPPENHOF K J, LAFARGE M W, MOESKOPS P, etal. Deformable image registration using convolutionalneural networks [J]. Proceedings of SPIE, 2018, 10574:105740S.
[10] NAZIB A, FOOKES C, PERRIN D. A comparative analysis of registration tools: Traditional vs deep learning approach on high resolution tissue cleared data [DB/OL]. (2018-10-18).https://arxiv.org/abs/1810.08315.
[11] BALAKRISHNAN G, ZHAO A, SABUNCU M R, etal. VoxelMorph: A learning framework for deformable medical image registration [J]. IEEE Transactions on Medical Imaging, 2019, 38(8): 2019Feb4.
[12] FU Y B, LEI Y, WANG T H, et al. LungRegNet: Anunsupervised deformable image registration methodfor 4D-CT lung [J]. Medical Physics, 2020, 47(4):1763-1774.
[13] HEINRICH M P, HANSEN L. Highly accurate and memory efficient unsupervised learning-based discreteCT registration using 2.5D displacement search [M]//Medical image computing and computer assisted intervention - MICCAI 2020. Cham: Springer, 2020:190-200.
[14] HERING A, HELDMANN S. Unsupervised learningfor large motion thoracic CT follow-up registration [J].Proceedings of SPIE, 2019, 10949: 109491B.
[15] LEE M C H, OKTAY O, SCHUH A, et al. Imageand-spatial transformer networks for structure-guided image registration [M]//Medical image computing and computer assisted intervention — MICCAI 2019.Cham: Springer, 2019: 337-345.
[16] SENTKER T, MADESTA F, WERNER R. GDLFIRE4D: Deep learning-based fast 4D CT image registration [M]//Medical image computing and computer assisted intervention — MICCAI 2018. Cham:Springer, 2018: 765-773.
[17] ZHAO S Y, DONG Y, CHANG E, et al. Recursivecascaded networks for unsupervised medical image registration [C]//2019 IEEE/CVF International Conference on Computer Vision. Seoul: IEEE, 2019: 10599-10609.
[18] DALCA A V, BALAKRISHNAN G, GUTTAG J, etal. Unsupervised learning for fast probabilistic diffeomorphic registration [M]//Medical image computingand computer assisted intervention — MICCAI 2018.Cham: Springer: 729-738.
[19] YANG X, KWITT R, STYNER M, et al. Quicksilver:Fast predictive image registration — A deep learningapproach [J]. NeuroImage, 2017, 158: 378-396.
[20] KREBS J, DELINGETTE H, MAILHE B, et al.Learning a probabilistic model for diffeomorphic registration [J]. IEEE Transactions on Medical Imaging,2019, 38(9): 2165-2176.
[21] MOK T C W, CHUNG A C S. Fast symmetric diffeomorphic image registration with convolutional neural networks [C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE,2020: 4643-4652.
[22] MOK T C W, CHUNG A C S. Large deformation diffeomorphic image registration with Laplacian pyramidnetworks [M]//Medical image computing and computer assisted intervention – MICCAI 2020. Cham:Springer, 2020: 211-221.
[23] HU R X, WANG H K, RISTANIEMI T, et al. LungCT image registration through landmark-constrainedlearning with convolutional neural network [C]//202042nd Annual International Conference of the IEEEEngineering in Medicine & Biology Society. Montreal:IEEE, 2020: 1368-1371.
[24] PUNN N S, AGARWAL S. Automated diagnosis ofCOVID-19 with limited posteroanterior chest X-ray images using fine-tuned deep neural networks [J]. Applied Intelligence, 2021, 51: 2689-2702.
[25] AMIRI M, BROOKS R, RIVAZ H. Fine-tuning U-net for ultrasound image segmentation: Different layers,different outcomes [J]. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 2020,67(12): 2510-2518.
[26] CHEN C, DOU Q, CHEN H, et al. Unsupervised bidirectional cross-modality adaptation via deeply synergistic image and feature alignment for medical imagesegmentation [J]. IEEE Transactions on Medical Imaging, 2020, 39(7): 2494-2505.
[27] FERRANTE E, OKTAY O, GLOCKER B, et al.On the adaptability of unsupervised CNN-based deformable image registration to unseen image domains[M]//Machine learning in medical imaging. Cham:Springer, 2018: 294-302.
[28] XUE P, DONG E Q, JI H Z. Lung 4D CT image registration based on high-order Markov random field [J].IEEE Transactions on Medical Imaging, 2020, 39(4):910-921.
[29] BALIK S, WEISS E, JAN N, et al. Evaluation of 4-dimensional computed tomography to 4-dimensionalcone-beam computed tomography deformable image registration for lung cancer adaptive radiation therapy [J]. International Journal of Radiation OncologyBiology Physics, 2013, 86(2): 372-379.
[30] CLARK K, VENDT B, SMITH K, et al. The cancer imaging archive (TCIA): Maintaining and operatinga public information repository [J]. Journal of Digital Imaging, 2013, 26(6): 1045-1057.
[31] HUGO G D, WEISS E, SLEEMAN W C, et al. A longitudinal four-dimensional computed tomography andcone beam computed tomography dataset for image-guided radiation therapy research in lung cancer [J].Medical Physics, 2017, 44(2): 762-771.
[32] ROMAN N O, SHEPHERD W, MUKHOPADHYAYN, et al. Interfractional positional variability of fiducial markers and primary tumors in locally advanced nonsmall-cell lung cancer during audiovisual biofeedback radiotherapy [J]. International Journal of RadiationOncology Biology Physics, 2012, 83(5): 1566-1572.
[33] DICE L R. Measures of the amount of ecologic association between species [J]. Ecology, 1945, 26(3): 297-302.