An unsupervised convolutional neural network-based algorithm for deformable image registration

The purpose of the work is to develop a deep unsupervised learning strategy for cone-beam CT (CBCT) to CT deformable image registration (DIR). This technique uses a deep convolutional inverse graphics network (DCIGN) based DIR algorithm implemented on 2 Nvidia 1080 Ti graphics processing units. The model is comprised of an encoding and decoding stage. The fully-convolutional encoding stage learns hierarchical features and simultaneously forms an information bottleneck, while the decoding stage restores the original dimensionality of the input image. Activations from the encoding stage are used as the input channels to a sparse DIR algorithm. DCIGN was trained using a distributive learning-based convolutional neural network architecture and used 285 head and neck patients to train, validate, and test the algorithm. The accuracy of the DCIGN algorithm was evaluated on 100 synthetic cases and 12 hold out test patient cases. The results indicate that DCIGN performed better than rigid registration, intensity corrected Demons, and landmark-guided deformable image registration for all evaluation metrics. DCIGN required ~14 h to train, and ~3.5 s to make a prediction on a 512  ×  512  ×  120 voxel image. In conclusion, DCIGN is able to maintain high accuracy in the presence of CBCT noise contamination, while simultaneously preserving high computational efficiency.

[1]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[2]  Linghong Zhou,et al.  CT to cone-beam CT deformable registration with simultaneous intensity correction , 2012, Medical physics.

[3]  Nilanjan Ray,et al.  Deep deformable registration: Enhancing accuracy by fully convolutional neural net , 2016, Pattern Recognit. Lett..

[4]  Nikos Paragios,et al.  DRAMMS: Deformable Registration via Attribute Matching and Mutual-Saliency Weighting , 2009, IPMI.

[5]  Junyi Xia,et al.  High performance computing for deformable image registration: Towards a new paradigm in adaptive radiotherapy. , 2008, Medical physics.

[6]  Yaoqin Xie,et al.  Auto-propagation of contours for adaptive prostate radiation therapy , 2008, Physics in medicine and biology.

[7]  Tom Vercauteren,et al.  Diffeomorphic demons: Efficient non-parametric image registration , 2009, NeuroImage.

[8]  Nikos Paragios,et al.  Handbook of Biomedical Imaging , 2015, Springer US.

[9]  Jonathan S. Maltz,et al.  Algorithm for X-ray Scatter, Beam-Hardening, and Beam Profile Correction in Diagnostic (Kilovoltage) and Treatment (Megavoltage) Cone Beam CT , 2008, IEEE Transactions on Medical Imaging.

[10]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[11]  W. McDavid,et al.  Deriving Hounsfield units using grey levels in cone beam CT: a clinical application. , 2012, Dento maxillo facial radiology.

[12]  Dinggang Shen,et al.  Learning Best Features and Deformation Statistics for Hierarchical Registration of MR Brain Images , 2007, IPMI.

[13]  Dinggang Shen,et al.  Feature‐based groupwise registration by hierarchical anatomical correspondence detection , 2012, Human brain mapping.

[14]  Weihua Mao,et al.  Canny edge-based deformable image registration , 2017, Physics in medicine and biology.

[15]  Dinggang Shen,et al.  HAMMER: hierarchical attribute matching mechanism for elastic registration , 2002, IEEE Transactions on Medical Imaging.

[16]  Josh Star-Lack,et al.  Correction for patient table-induced scattered radiation in cone-beam computed tomography (CBCT). , 2011, Medical physics.

[17]  Joshua B. Tenenbaum,et al.  Deep Convolutional Inverse Graphics Network , 2015, NIPS.

[18]  N. Ayache,et al.  Statistics on Diffeomorphisms in a Log-Euclidean Framework , 2006 .

[19]  Carl-Fredrik Westin,et al.  Improving Registration Using Multi-channel Diffeomorphic Demons Combined with Certainty Maps , 2011, MBIA.

[20]  L. Xing,et al.  Feature-based rectal contour propagation from planning CT to cone beam CT. , 2008, Medical physics.

[21]  Eduard Schreibmann,et al.  Quantitative evaluation of a cone‐beam computed tomography–planning computed tomography deformable image registration method for adaptive radiation therapy , 2007, Journal of applied clinical medical physics.

[22]  Vasant Kearney,et al.  CyberArc: a non-coplanar-arc optimization algorithm for CyberKnife , 2017, Physics in medicine and biology.

[23]  Yaozong Gao,et al.  Learning‐based deformable image registration for infant MR images in the first year of life , 2017, Medical physics.

[24]  Dinggang Shen,et al.  Learning-based deformable registration of MR brain images , 2006, IEEE Transactions on Medical Imaging.

[25]  Jing Wang,et al.  Penalized weighted least-squares approach to sinogram noise reduction and image reconstruction for low-dose X-ray computed tomography , 2006, IEEE Transactions on Medical Imaging.

[26]  Dinggang Shen,et al.  Deformable segmentation of 3-D ultrasound prostate images using statistical texture matching method , 2006, IEEE Transactions on Medical Imaging.

[27]  Benson Mwangi,et al.  A Review of Feature Reduction Techniques in Neuroimaging , 2013, Neuroinformatics.

[28]  Torsten Rohlfing,et al.  Image Similarity and Tissue Overlaps as Surrogates for Image Registration Accuracy: Widely Used but Unreliable , 2012, IEEE Transactions on Medical Imaging.

[29]  Sanford L. Meeks,et al.  Benchmarking of five commercial deformable image registration algorithms for head and neck patients , 2016, Journal of applied clinical medical physics.

[30]  Nicholas Ayache,et al.  Three-dimensional multimodal brain warping using the Demons algorithm and adaptive intensity corrections , 2001, IEEE Transactions on Medical Imaging.

[31]  Heung-Il Suk,et al.  Deep Learning in Medical Image Analysis. , 2017, Annual review of biomedical engineering.

[32]  Hongbing Lu,et al.  Nonlinear sinogram smoothing for low-dose X-ray CT , 2004 .

[33]  Dinggang Shen,et al.  Scalable High-Performance Image Registration Framework by Unsupervised Deep Feature Representations Learning , 2016, IEEE Transactions on Biomedical Engineering.

[34]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[35]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[36]  Xuejun Gu,et al.  Automated landmark-guided deformable image registration , 2015, Physics in medicine and biology.

[37]  Lei Zhu,et al.  Scatter correction for cone-beam CT in radiation therapy. , 2009 .

[38]  David Zhang,et al.  FSIM: A Feature Similarity Index for Image Quality Assessment , 2011, IEEE Transactions on Image Processing.

[39]  J. Dinten,et al.  A new method for x-ray scatter correction: first assessment on a cone-beam CT experimental setup , 2007, Physics in medicine and biology.

[40]  Boyd McCurdy,et al.  Cone beam computerized tomography: the effect of calibration of the Hounsfield unit number to electron density on dose calculation accuracy for adaptive radiation therapy , 2009, Physics in medicine and biology.

[41]  Lei Xing,et al.  Evaluation of on-board kV cone beam CT (CBCT)-based dose calculation , 2007, Physics in medicine and biology.

[42]  Lei Zhu,et al.  Scatter correction for full-fan volumetric CT using a stationary beam blocker in a single full scan. , 2011, Medical physics.

[43]  Fang-Fang Yin,et al.  Dosimetric feasibility of cone-beam CT-based treatment planning compared to CT-based treatment planning. , 2006, International journal of radiation oncology, biology, physics.

[44]  Mariana Guerrero,et al.  Deformable planning CT to cone-beam CT image registration in head-and-neck cancer. , 2011, Medical physics.

[45]  John Ashburner,et al.  A fast diffeomorphic image registration algorithm , 2007, NeuroImage.

[46]  David G. Lowe,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004, International Journal of Computer Vision.