Wenyang Liu

A continuous surface reconstruction method on point cloud captured from a 3D surface photogrammetry system.

PURPOSE To accurately and efficiently reconstruct a continuous surface from noisy point clouds captured by a surface photogrammetry system (VisionRT). METHODS The authors have developed a level-set based surface reconstruction method on point clouds captured by a surface photogrammetry system (VisionRT). The proposed method reconstructs an implicit and continuous representation of the underlying patient surface by optimizing a regularized fitting energy, offering extra robustness to noise and missing measurements. By contrast to explicit/discrete meshing-type schemes, their continuous representation is particularly advantageous for subsequent surface registration and motion tracking by eliminating the need for maintaining explicit point correspondences as in discrete models. The authors solve the proposed method with an efficient narrowband evolving scheme. The authors evaluated the proposed method on both phantom and human subject data with two sets of complementary experiments. In the first set of experiment, the authors generated a series of surfaces each with different black patches placed on one chest phantom. The resulting VisionRT measurements from the patched area had different degree of noise and missing levels, since VisionRT has difficulties in detecting dark surfaces. The authors applied the proposed method to point clouds acquired under these different configurations, and quantitatively evaluated reconstructed surfaces by comparing against a high-quality reference surface with respect to root mean squared error (RMSE). In the second set of experiment, the authors applied their method to 100 clinical point clouds acquired from one human subject. In the absence of ground-truth, the authors qualitatively validated reconstructed surfaces by comparing the local geometry, specifically mean curvature distributions, against that of the surface extracted from a high-quality CT obtained from the same patient. RESULTS On phantom point clouds, their method achieved submillimeter reconstruction RMSE under different configurations, demonstrating quantitatively the faith of the proposed method in preserving local structural properties of the underlying surface in the presence of noise and missing measurements, and its robustness toward variations of such characteristics. On point clouds from the human subject, the proposed method successfully reconstructed all patient surfaces, filling regions where raw point coordinate readings were missing. Within two comparable regions of interest in the chest area, similar mean curvature distributions were acquired from both their reconstructed surface and CT surface, with mean and standard deviation of (μrecon=-2.7×10(-3) mm(-1), σrecon=7.0×10(-3) mm(-1)) and (μCT=-2.5×10(-3) mm(-1), σCT=5.3×10(-3) mm(-1)), respectively. The agreement of local geometry properties between the reconstructed surfaces and the CT surface demonstrated the ability of the proposed method in faithfully representing the underlying patient surface. CONCLUSIONS The authors have integrated and developed an accurate level-set based continuous surface reconstruction method on point clouds acquired by a 3D surface photogrammetry system. The proposed method has generated a continuous representation of the underlying phantom and patient surfaces with good robustness against noise and missing measurements. It serves as an important first step for further development of motion tracking methods during radiotherapy.

Estimating nonrigid motion from inconsistent intensity with robust shape features.

PURPOSE To develop a nonrigid motion estimation method that is robust to heterogeneous intensity inconsistencies amongst the image pairs or image sequence. METHODS Intensity and contrast variations, as in dynamic contrast enhanced magnetic resonance imaging, present a considerable challenge to registration methods based on general discrepancy metrics. In this study, the authors propose and validate a novel method that is robust to such variations by utilizing shape features. The geometry of interest (GOI) is represented with a flexible zero level set, segmented via well-behaved regularized optimization. The optimization energy drives the zero level set to high image gradient regions, and regularizes it with area and curvature priors. The resulting shape exhibits high consistency even in the presence of intensity or contrast variations. Subsequently, a multiscale nonrigid registration is performed to seek a regular deformation field that minimizes shape discrepancy in the vicinity of GOIs. RESULTS To establish the working principle, realistic 2D and 3D images were subject to simulated nonrigid motion and synthetic intensity variations, so as to enable quantitative evaluation of registration performance. The proposed method was benchmarked against three alternative registration approaches, specifically, optical flow, B-spline based mutual information, and multimodality demons. When intensity consistency was satisfied, all methods had comparable registration accuracy for the GOIs. When intensities among registration pairs were inconsistent, however, the proposed method yielded pronounced improvement in registration accuracy, with an approximate fivefold reduction in mean absolute error (MAE = 2.25 mm, SD = 0.98 mm), compared to optical flow (MAE = 9.23 mm, SD = 5.36 mm), B-spline based mutual information (MAE = 9.57 mm, SD = 8.74 mm) and mutimodality demons (MAE = 10.07 mm, SD = 4.03 mm). Applying the proposed method on a real MR image sequence also provided qualitatively appealing results, demonstrating good feasibility and applicability of the proposed method. CONCLUSIONS The authors have developed a novel method to estimate the nonrigid motion of GOIs in the presence of spatial intensity and contrast variations, taking advantage of robust shape features. Quantitative analysis and qualitative evaluation demonstrated good promise of the proposed method. Further clinical assessment and validation is being performed.

Respiratory motion prediction and prospective correction for free‐breathing arterial spin‐labeled perfusion MRI of the kidneys

Purpose: Respiratory motion prediction using an artificial neural network (ANN) was integrated with pseudocontinuous arterial spin labeling (pCASL) MRI to allow free‐breathing perfusion measurements in the kidney. In this study, we evaluated the performance of the ANN to accurately predict the location of the kidneys during image acquisition. Methods: A pencil‐beam navigator was integrated with a pCASL sequence to measure lung/diaphragm motion during ANN training and the pCASL transit delay. The ANN algorithm ran concurrently in the background to predict organ location during the 0.7‐s 15‐slice acquisition based on the navigator data. The predictions were supplied to the pulse sequence to prospectively adjust the axial slice acquisition to match the predicted organ location. Additional navigators were acquired immediately after the multislice acquisition to assess the performance and accuracy of the ANN. The technique was tested in eight healthy volunteers. Results: The root‐mean‐square error (RMSE) and mean absolute error (MAE) for the eight volunteers were 1.91 ± 0.17 mm and 1.43 ± 0.17 mm, respectively, for the ANN. The RMSE increased with transit delay. The MAE typically increased from the first to last prediction in the image acquisition. The overshoot was 23.58% ± 3.05% using the target prediction accuracy of ± 1 mm. Conclusion: Respiratory motion prediction with prospective motion correction was successfully demonstrated for free‐breathing perfusion MRI of the kidney. The method serves as an alternative to multiple breathholds and requires minimal effort from the patient.

A robust real‐time surface reconstruction method on point clouds captured from a 3D surface photogrammetry system

PURPOSE To develop a robust and real-time surface reconstruction method on point clouds captured from a 3D surface photogrammetry system. METHODS The authors have developed a robust and fast surface reconstruction method on point clouds acquired by the photogrammetry system, without explicitly solving the partial differential equation required by a typical variational approach. Taking advantage of the overcomplete nature of the acquired point clouds, their method solves and propagates a sparse linear relationship from the point cloud manifold to the surface manifold, assuming both manifolds share similar local geometry. With relatively consistent point cloud acquisitions, the authors propose a sparse regression (SR) model to directly approximate the target point cloud as a sparse linear combination from the training set, assuming that the point correspondences built by the iterative closest point (ICP) is reasonably accurate and have residual errors following a Gaussian distribution. To accommodate changing noise levels and/or presence of inconsistent occlusions during the acquisition, the authors further propose a modified sparse regression (MSR) model to model the potentially large and sparse error built by ICP with a Laplacian prior. The authors evaluated the proposed method on both clinical point clouds acquired under consistent acquisition conditions and on point clouds with inconsistent occlusions. The authors quantitatively evaluated the reconstruction performance with respect to root-mean-squared-error, by comparing its reconstruction results against that from the variational method. RESULTS On clinical point clouds, both the SR and MSR models have achieved sub-millimeter reconstruction accuracy and reduced the reconstruction time by two orders of magnitude to a subsecond reconstruction time. On point clouds with inconsistent occlusions, the MSR model has demonstrated its advantage in achieving consistent and robust performance despite the introduced occlusions. CONCLUSIONS The authors have developed a fast and robust surface reconstruction method on point clouds captured from a 3D surface photogrammetry system, with demonstrated sub-millimeter reconstruction accuracy and subsecond reconstruction time. It is suitable for real-time motion tracking in radiotherapy, with clear surface structures for better quantifications.

Segmentation with a shape dictionary

Image segmentation plays an important role in many medical applications. Automatic segmentation algorithms are challenged by low SNR and significant artifacts resulting from motion and signal voids. In this study, we propose a novel level set based segmentation method with a shape dictionary. Unlike previous studies that use a single template or probabilistic models, we propose to construct a shape dictionary and model the shape prior as sparse combinations of shape templates in the dictionary. The proposed method generated promising segmentation results on low SNR MR images, even with signal voids.

Real-time motion estimation with MRI

Real-time motion estimation is challenging, especially with MR imaging where an intrinsic tradeoff exists between spatial and temporal resolution. The spatially varying pharmacokinetic behavior of contrast agents contributes to intensity inhomogeneity and contrast changing, introducing a further obstacle in motion estimation. In this study, we propose a novel level set based motion estimation method based on the observation that higher order geometric features present in anatomic boundaries are less influenced by contrast and intensity inhomogeneity due to low SNR and/or contrast changes. More specifically, our method first estimates the movement of anatomical boundaries, and then extrapolates them to the whole domain of interest. Preliminary tests have demonstrated desirable error statistics on MR images with simulated contrast and intensity inhomogeneity, as would be the case when contrast agents are used. Tests based on real-time MR image sequences have revealed visually appealing results, conforming to prior physical and physiological understanding.

Free-breathing perfusion MRI using multislice pCASL

Arterial spin labeling (ASL) is a difference imaging method that is sensitive to motion between successive label and control acquisitions, restricting its use in the abdomen. Respiratory motion prediction (RMP) using an artificial neutral network (ANN) and navigator echoes was developed to enable free-breathing perfusion measurements of the kidney using pseudo-continuous arterial spin labeling (pCASL) magnetic resonance imaging (MRI).

TU-CD-BRA-03: Left Ventricle Segmentation with a Coupled Length Regularization and Sparse Shape Prior

Purpose: To perform accurate segmentation on left ventricles from Cine MR images Methods: We have developed a novel variational segmentation method that incorporates prior knowledge on both geometrical relations and shapes of the endocardium and epicardium. In contrast to conventional approaches that preserve a constant distance between the endo- and epicardial contours, we propose to maintain a smoothly varying distance between the endo- and epicardium represented by two separate level set functions, which is more robust to shape variations across slices and phases among different patients. The coupled level set representation further allows us to incorporate a sparse composite shape prior to boost the performance. A robust data fidelity of Gaussian mixture is utilized to represent overlapped intensity distributions of each cardiac region under the condition of insufficient gradient information. We evaluated the proposed method on datasets from MICCAI left ventricle segmentation challenge, and compared our method against other state-of-the-art approaches based on Dice similarity coefficient (DSC). Results: Our method successfully segmented all left ventricles in 15 validation datasets from patients of different pathologies. It achieved competitive if not better DSC accuracy compared to other state-of-the-art methods, with mean DSC of 0.89 and standard deviation of 0.04 on endocardium segmentation, and mean DSC of 0.94 and standard deviation of 0.01 on epicardium segmentation. Conclusion: Introducing anatomy-specific geometric coupling and sparse composite shape priors on endo- and epicardium, in addition to length regularization, into the variational method has demonstrated its advantage in the challenging problem of left ventricle segmentation. A joint left and right ventricle segmentation method is under development.

TU‐F‐BRF‐05: Level Set Based Segmentation with a Dynamic Shape Prior

PURPOSE To perform accurate segmentation on low-SNR MR images subject to motion artifacts and signal voids. METHODS We have developed a novel level set based segmentation method that incorporates a dynamic shape prior. In contrast to conventional shape prior models that are either based on a single template or statistical models, we modeled the shape prior as sparse linear combinations of templates in a shape library. The proposed method was applied to segment real-time kidney geometry from an abdominal MR series acquired under respiration and compared with the Chan-Vese approach. Synthetic occlusions were further introduced to assess performance robustness. We compared the segmentation results with the ground truth contours and evaluated the segmentation accuracy based on the Dice similarity coefficient. RESULTS The proposed method successfully segmented all kidneys in the testing series. The corresponding shape priors were reasonably reconstructed from a small number of shape templates in the library, as expected. On testing dataset without occlusion, the proposed method achieved mean DSC of 0.97, which was more accurate than the Chan-Vese approach with a DSC of 0.93. When tested on dataset with synthetic occlusions, the proposed method made robust inference on the occlusion sites and achieved mean DSC of 0.95, compared to 0.76 from the Chan-Vese method, demonstrating the advantage of incorporating the proposed shape prior. CONCLUSION We have developed a novel level set based segmentation method with a novel regularization to incorporate a dynamic shape prior. The shape prior is dynamically updated during the segmentation process as a sparse linear combination of templates from a shape library. Preliminary results have demonstrated the ability of the proposed method in improving segmentation accuracy, especially when noises and/or signal voids are present. Future work will consider dynamically update and evolve the shape library to encode shape priors as segmentations are being performed.

SU‐C‐17A‐06: Motion Compensation in Dynamic Contrast Enhanced MRI

PURPOSE To apply a newly developed shape-based registration scheme for motion compensation in MR urography and verify its efficacy in facilitating quantitative functional analysis. METHODS We have recently developed a shape-based registration scheme that is robust w.r.t. intensity inconsistency. In this study, we utilized this robust registration tool to estimate kidney motion during the MRU scans and compensate for such motion to facilitate the quantitative functional analysis. To validate the efficacy of this scheme, MRU analysis was performed on dataset acquired from sedated subjects to obtain ground-truth (motion-free) functional estimates. Physiologically sound motion was then simulated to synthesize image sequences influenced by respiratory motion. Quantitative assessment and comparison were performed amongst ground-truth, calculations without and with motion compensation for the following set of functional parameters: the contrast dynamic of the left and right cortex, medulla and aorta, the Patlak number and the globular filtration rate (GFR) based on the Patalk-Rutland model, and the Patlak differential renal function (pDRF). RESULTS Without motion compensation, the generated relative enhancement curves contained large fluctuations and the estimated GFR values were underestimated by 26% (75.4 ml/min) and 30% (95.3 ml/min) compared with the ground truth of 101.9ml/min and 134.3ml/min for the left and right kidney respectively. Such large errors could result in misleading diagnosis if a typical threshold of 90ml/min were used to determine renal function abnormality. Upon proposed motion compensation, the relative enhancement curves were much smoother and the GFR estimation errors for the left and right kidneys reduced to 0.4% and 1.6% respectively, demonstrating the advantage of the introduced motion compensation method. CONCLUSION The developed motion compensation method has demonstrated its ability to facilitate quantitative MRU functional analysis, with improved accuracy of pharmacokinetic modeling and quantitative parameter estimations. Future work will consider incorporating more complex and realistic pharmacokinetic models into the MRU analysis.