Deformable lung 4DCT image registration via landmark-driven cycle network.
deep learning
deformable image registration
lung CT
radiotherapy
Journal
Medical physics
ISSN: 2473-4209
Titre abrégé: Med Phys
Pays: United States
ID NLM: 0425746
Informations de publication
Date de publication:
14 Sep 2023
14 Sep 2023
Historique:
received:
16
08
2022
accepted:
01
09
2023
medline:
14
9
2023
pubmed:
14
9
2023
entrez:
14
9
2023
Statut:
aheadofprint
Résumé
An automated, accurate, and efficient lung four-dimensional computed tomography (4DCT) image registration method is clinically important to quantify respiratory motion for optimal motion management. The purpose of this work is to develop a weakly supervised deep learning method for 4DCT lung deformable image registration (DIR). The landmark-driven cycle network is proposed as a deep learning platform that performs DIR of individual phase datasets in a simulation 4DCT. This proposed network comprises a generator and a discriminator. The generator accepts moving and target CTs as input and outputs the deformation vector fields (DVFs) to match the two CTs. It is optimized during both forward and backward paths to enhance the bi-directionality of DVF generation. Further, the landmarks are used to weakly supervise the generator network. Landmark-driven loss is used to guide the generator's training. The discriminator then judges the realism of the deformed CT to provide extra DVF regularization. We performed four-fold cross-validation on 10 4DCT datasets from the public DIR-Lab dataset and a hold-out test on our clinic dataset, which included 50 4DCT datasets. The DIR-Lab dataset was used to evaluate the performance of the proposed method against other methods in the literature by calculating the DIR-Lab Target Registration Error (TRE). The proposed method outperformed other deep learning-based methods on the DIR-Lab datasets in terms of TRE. Bi-directional and landmark-driven loss were shown to be effective for obtaining high registration accuracy. The mean and standard deviation of TRE for the DIR-Lab datasets was 1.20 ± 0.72 mm and the mean absolute error (MAE) and structural similarity index (SSIM) for our datasets were 32.1 ± 11.6 HU and 0.979 ± 0.011, respectively. The landmark-driven cycle network has been validated and tested for automatic deformable image registration of patients' lung 4DCTs with results comparable to or better than competing methods.
Sections du résumé
BACKGROUND
BACKGROUND
An automated, accurate, and efficient lung four-dimensional computed tomography (4DCT) image registration method is clinically important to quantify respiratory motion for optimal motion management.
PURPOSE
OBJECTIVE
The purpose of this work is to develop a weakly supervised deep learning method for 4DCT lung deformable image registration (DIR).
METHODS
METHODS
The landmark-driven cycle network is proposed as a deep learning platform that performs DIR of individual phase datasets in a simulation 4DCT. This proposed network comprises a generator and a discriminator. The generator accepts moving and target CTs as input and outputs the deformation vector fields (DVFs) to match the two CTs. It is optimized during both forward and backward paths to enhance the bi-directionality of DVF generation. Further, the landmarks are used to weakly supervise the generator network. Landmark-driven loss is used to guide the generator's training. The discriminator then judges the realism of the deformed CT to provide extra DVF regularization.
RESULTS
RESULTS
We performed four-fold cross-validation on 10 4DCT datasets from the public DIR-Lab dataset and a hold-out test on our clinic dataset, which included 50 4DCT datasets. The DIR-Lab dataset was used to evaluate the performance of the proposed method against other methods in the literature by calculating the DIR-Lab Target Registration Error (TRE). The proposed method outperformed other deep learning-based methods on the DIR-Lab datasets in terms of TRE. Bi-directional and landmark-driven loss were shown to be effective for obtaining high registration accuracy. The mean and standard deviation of TRE for the DIR-Lab datasets was 1.20 ± 0.72 mm and the mean absolute error (MAE) and structural similarity index (SSIM) for our datasets were 32.1 ± 11.6 HU and 0.979 ± 0.011, respectively.
CONCLUSION
CONCLUSIONS
The landmark-driven cycle network has been validated and tested for automatic deformable image registration of patients' lung 4DCTs with results comparable to or better than competing methods.
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Subventions
Organisme : NIBIB NIH HHS
ID : R56 EB033332
Pays : United States
Organisme : NIH HHS
ID : R01CA215718
Pays : United States
Organisme : NCI NIH HHS
ID : R01 CA272991
Pays : United States
Organisme : NIH HHS
ID : R01CA272991
Pays : United States
Organisme : NIH HHS
ID : R01EB032680
Pays : United States
Organisme : NIH HHS
ID : R56EB033332
Pays : United States
Informations de copyright
© 2023 American Association of Physicists in Medicine.
Références
Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer statistics, 2021. CA: A Cancer J Clin. 2021;71(1):7-33.
Miller KD, Nogueira L, Mariotto AB, et al. Cancer treatment and survivorship statistics, 2019. CA: A Cancer J Clin. 2019;69(5):363-385.
Keall PJ, Mageras GS, Balter JM, et al. The management of respiratory motion in radiation oncology report of AAPM Task Group 76. Med Phys. 2006;33(10):3874-3900. [published online ahead of print 2006/11/09].
Benedict SH, Yenice KM, Followill D, et al. Stereotactic body radiation therapy: the report of AAPM Task Group 101. Med Phys. 2010;37(8):4078-4101. [published online ahead of print 2010/10/01].
Keall P. 4-dimensional computed tomography imaging and treatment planning. Semin Radiat Oncol. 2004;14(1):81-90. [published online ahead of print 2004/01/31].
Speight R, Sykes J, Lindsay R, Franks K, Thwaites D. The evaluation of a deformable image registration segmentation technique for semi-automating internal target volume (ITV) production from 4DCT images of lung stereotactic body radiotherapy (SBRT) patients. Radiother Oncol. 2011;98(2):277-283. [published online ahead of print 2011/01/25].
Rong Y, Rosu-Bubulac M, Benedict SH, et al. Rigid and deformable image registration for radiation therapy: a self-study evaluation guide for NRG Oncology Clinical Trial Participation. Pract Radiat Oncol. 2021;11(4):282-298. [published online ahead of print 2021/03/05].
Tong Y, Yin Y, Cheng P, Gong G. Impact of deformable image registration on dose accumulation applied electrocardiograph-gated 4DCT in the heart and left ventricular myocardium during esophageal cancer radiotherapy. Radiat Oncol. 2018;13(1):145. [published online ahead of print 2018/08/12].
Hegi-Johnson F, Keall P, Barber J, Bui C, Kipritidis J. Evaluating the accuracy of 4D-CT ventilation imaging: first comparison with Technegas SPECT ventilation. Med Phys. 2017;44(8):4045-4055. [published online ahead of print 2017/05/10].
Fu Y, Lei Y, Wang T, Curran WJ, Liu T, Yang X. Deep learning in medical image registration: a review. Phys Med Biol. 2020;65(20):20TR01.
Haskins G, Kruger U, Yan P. Deep learning in medical image registration: a survey. Machine Vis Appl. 2020;31(1):8.
Jaderberg M, Simonyan K, Zisserman A, Kavukcuoglu K. Spatial Transformer Networks. 2015. Accessed June 01, 2015. doi: 10.48550/arXiv.1506.02025:arXiv:1506.02025. https://ui.adsabs.harvard.edu/abs/2015arXiv150602025J
de Vos BD, Berendsen FF, Viergever MA, Sokooti H, Staring M, Išgum I. A deep learning framework for unsupervised affine and deformable image registration. Med Image Anal. 2019;52:128-143. [published online ahead of print 2018/12/24].
Sentker T, Madesta F, Werner R. GDL-FIRE4D: deep Learning-Based Fast 4D CT Image Registration. Medical Image Computing and Computer Assisted Intervention-MICCAI 2018. Cham; 2018.
Eppenhof KAJ, Pluim JPW. Pulmonary CT registration through supervised learning with convolutional neural networks. IEEE Trans Med Imaging. 2019;38(5):1097-1105. [published online ahead of print 2018/10/30].
Fu Y, Lei Y, Wang T, et al. LungRegNet: an unsupervised deformable image registration method for 4D-CT lung. Med Phys. 2020;47(4):1763-1774. [published online ahead of print 2020/02/06].
Fu Y, Wang T, Lei Y, et al. Deformable MR-CBCT prostate registration using biomechanically constrained deep learning networks. Med Phys. 2021;48(1):253-263.
Lei Y, Fu Y, Wang T, et al. 4D-CT deformable image registration using multiscale unsupervised deep learning. Phys Med Biol. 2020;65(8):085003.
Zhang Y, Wu X, Gach HM, Li H, Yang D. GroupRegNet: a groupwise one-shot deep learning-based 4D image registration method. Phys Med Biol. 2021;66(4):045030. [published online ahead of print 20210212].
Kim B, Kim DH, Park SH, Kim J, Lee J-G, Ye JC. CycleMorph: cycle consistent unsupervised deformable image registration. Med Image Analysis. 2021;71:102036.
Zhang J. Inverse-consistent deep networks for unsupervised deformable image registration. ArXiv. 2018:abs/1809.03443.
Li H, Fan Y. Non-rigid image registration using self-supervised fully convolutional networks without training data. Proc IEEE Int Symp Biomed Imaging. 2018;2018:1075-1078. [published online ahead of print 05/24].
Reuter M, Rosas HD, Fischl B. Highly accurate inverse consistent registration: a robust approach. Neuroimage. 2010;53(4):1181-1196. [published online ahead of print 20100714].
Zhang Y, He X, Tian Z, et al. Multi-needle detection in 3D ultrasound images using unsupervised Order-Graph regularized sparse dictionary learning. IEEE Trans Med Imaging. 2020;39(7):2302-2315. [published online ahead of print 20200122].
Castillo E, Castillo R, Martinez J, Shenoy M, Guerrero T. Four-dimensional deformable image registration using trajectory modeling. Phys Med Biol. 2010;55(1):305-327. [published online ahead of print 2009/12/17].
Castillo R, Castillo E, Guerra R, et al. A framework for evaluation of deformable image registration spatial accuracy using large landmark point sets. Phys Med Biol. 2009;54(7):1849-1870. [published online ahead of print 2009/03/07].
Fu Y, Wu X, Thomas AM, Li HH, Yang D. Automatic large quantity landmark pairs detection in 4DCT lung images. Med Phys. 2019;46(10):4490-4501. [published online ahead of print 2019/07/19].
Lei Y, Fu Y, Tian Z, et al. Deformable CT image registration via a dual feasible neural network. Med Phys. 2022. doi:10.1002/mp.15875. [published online ahead of print 2022/07/24].
Lee S, Yan G, Lu B, Kahler D, Li JG, Sanjiv SS. Impact of scanning parameters and breathing patterns on image quality and accuracy of tumor motion reconstruction in 4D CBCT: a phantom study. J Appl Clin Med Phys. 2015;16(6):195-212.
Watkins WT, Li R, Lewis J, et al. Patient-specific motion artifacts in 4DCT. Med Phys. 2010;37(6Part1):2855-2861.