Image Registration with Sliding Motion Constraints for 4D CT Motion Correction

Derksen, Alexander; Heldmann, Stefan; Polzin, Thomas; Berkels, Benjamin

doi:10.1007/978-3-662-46224-9_58

Cited by 7 publications

(8 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To address this problem, several approaches have been proposed, which consider an initial segmentation of moving and less-moving regions in the image to restrict the regularization at the boundaries of the sliding areas. [7][8][9][10][11] As an example, Delmon et al 9 have proposed a B-spline-based registration method such that regularization is enforced along the direction tangential to the segmented boundary and it is relaxed in the normal direction. Hua et al 11 have extended the B-spline approach by incorporating an additional term that acts on a subset of control points, for which the boundary intersects the support of their corresponding basis function.…”

Section: Deformable Image Registration and Motion Maskmentioning

confidence: 99%

“…Also a finite-element-based approach has been proposed to perform lung registration preserving motion discontinuity. 10 Although such methods often use presegmentation of the lungs alone, Vandemeulebroucke et al 12 have developed a so-called motion mask that encompasses lungs and viscera, and thus delineates a region, in which the motion field is supposed to be smooth (Figure 1). Its boundary defines where slidingand thus discontinuity in the vector field-occur between the rib cage and the internal structures, when breathing.…”

Section: Deformable Image Registration and Motion Maskmentioning

confidence: 99%

“…To address this problem, several approaches have been proposed, which consider an initial segmentation of moving and less‐moving regions in the image to restrict the regularization at the boundaries of the sliding areas 7–11 . As an example, Delmon et al 9 .…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Improving motion‐mask segmentation in thoracic CT with multiplanar U‐nets

et al. 2021

View full text Add to dashboard Cite

Purpose. Motion-mask segmentation from thoracic CT images is the process of extracting the region that encompasses lungs and viscera, where large displacements occur during breathing. It has been shown to help image registration between different respiratory phases. This registration step is, for example, useful for radiotherapy planning or calculating local lung ventilation. Knowing the location of motion discontinuity, i.e., sliding motion near the pleura, allows a better control of the registration preventing unrealistic estimates. Nevertheless, existing methods for motion-mask segmentation are not robust enough to be used in clinical routine. This article shows that it is feasible to overcome this lack of robustness by using a lightweight deep-learning approach usable on a standard computer, and this even without data augmentation or advanced model design.Methods. A convolutional neural-network architecture with three 2D U-nets for the three main orientations (sagittal, coronal, axial) was proposed. Predictions generated by the three Unets were combined by majority voting to provide a single 3D segmentation of the motion mask.The networks were trained on a database of non-small cell lung cancer 4D CT images of 43 patients.Training and evaluation were done with a K-fold cross-validation strategy. Evaluation was based on a visual grading by two experts according to the appropriateness of the segmented motion mask for the registration task, and on a comparison with motion masks obtained by a baseline method using level sets. A second database (76 CT images of patients with early-stage COVID-19), unseen during training, was used to assess the generalizability of the trained neural network.Results. The proposed approach outperformed the baseline method in terms of quality and robustness: the success rate increased from 53% to 79% without producing any failure. It also achieved a speed-up factor of 60 with GPU, or 17 with CPU. The memory footprint was low: less than 5 GB GPU RAM for training and less than 1 GB GPU RAM for inference. When evaluated on a dataset with images differing by several characteristics (CT device, pathology, and field of view), the proposed method improved the success rate from 53% to 83%. Conclusion.With 5-second processing time on a mid-range GPU and success rates around 80%, the proposed approach seems fast and robust enough to be routinely used in clinical practice.The success rate can be further improved by incorporating more diversity in training data via data augmentation and additional annotated images from different scanners and diseases. The code and trained model are publicly available.

show abstract

Section: Deformable Image Registration and Motion Maskmentioning

confidence: 99%

Section: Deformable Image Registration and Motion Maskmentioning

confidence: 99%

See 1 more Smart Citation

Improving motion‐mask segmentation in thoracic CT with multiplanar U‐nets

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Exemplary early applications of TV regularization for optical flow estimation can be found in [25,35] and for image registration in [28,34]. In the context of medical image processing, TV regularization is particularly interesting when modeling non-smooth sliding motions, since it eliminates the necessity to explicitly mask all sliding interfaces beforehand [10].…”

Section: Total Variation Regularizationmentioning

confidence: 99%

Deformable Groupwise Image Registration using Low-Rank and Sparse Decomposition

Haase¹,

Heldmann²,

Lellmann³

2020

Preprint

Self Cite

View full text Add to dashboard Cite

Low-rank and sparse decompositions and robust PCA (RPCA) are highly successful techniques in image processing and have recently found use in groupwise image registration. In this paper, we investigate the drawbacks of the most common RPCA-dissimilarity metric in image registration and derive an improved version. In particular, this new metric models low-rank requirements through explicit constraints instead of penalties and thus avoids the pitfalls of the established metric. Equipped with total variation regularization, we present a theoretically justified multilevel scheme based on first-order primal-dual optimization to solve the resulting non-parametric registration problem. As confirmed by numerical experiments, our metric especially lends itself to data involving recurring changes in object appearance and potential sparse perturbations. We numerically compare its peformance to a number of related approaches.

show abstract

“…To our knowledge, a nonaddressed problem is the non‐rigid misalignments (NRM) in ET in materials science. Yet, non‐rigid registration has been widely used for medical applications (Hill et al ., ; Chui & Rangarajan, ; Derksen et al ., ). Recent works even propose to use nonrigid registration for STEM images (Berkels et al ., ; Jones et al ., ).…”

Section: Introductionmentioning

confidence: 97%

Non‐rigid alignment in electron tomography in materials science

Printemps

Bernier

Bleuet

et al. 2016

Journal of Microscopy

View full text Add to dashboard Cite

Electron tomography is a key technique that enables the visualization of an object in three dimensions with a resolution of about a nanometre. High-quality 3D reconstruction is possible thanks to the latest compressed sensing algorithms and/or better alignment and preprocessing of the 2D projections. Rigid alignment of 2D projections is routine in electron tomography. However, it cannot correct misalignments induced by (i) deformations of the sample due to radiation damage or (ii) drifting of the sample during the acquisition of an image in scanning transmission electron microscope mode. In both cases, those misalignments can give rise to artefacts in the reconstruction. We propose a simple-to-implement non-rigid alignment technique to correct those artefacts. This technique is particularly suited for needle-shaped samples in materials science. It is initiated by a rigid alignment of the projections and it is then followed by several rigid alignments of different parts of the projections. Piecewise linear deformations are applied to each projection to force them to simultaneously satisfy the rigid alignments of the different parts. The efficiency of this technique is demonstrated on three samples, an intermetallic sample with deformation misalignments due to a high electron dose typical to spectroscopic electron tomography, a porous silicon sample with an extremely thin end particularly sensitive to electron beam and another porous silicon sample that was drifting during image acquisitions.

show abstract

Image Registration with Sliding Motion Constraints for 4D CT Motion Correction

Cited by 7 publications

References 7 publications

Improving motion‐mask segmentation in thoracic CT with multiplanar U‐nets

Improving motion‐mask segmentation in thoracic CT with multiplanar U‐nets

Deformable Groupwise Image Registration using Low-Rank and Sparse Decomposition

Non‐rigid alignment in electron tomography in materials science

Contact Info

Product

Resources

About