Automatic model-based segmentation of the heart in CT images

A novel and rapid approach to estimate patient-specific distortions based on mDIXON MRI

While MRI-only radiation treatment planning (RTP) is becoming more widespread, a robust clinical solution for patient-specific distortion corrections is not available. This work explores B ₀ mapping based on mDIXON imaging, often performed for MR-only RTP, as an alternative to separate dual-acquisition gradient-recalled echo imaging, with the overarching goal of developing an efficient and robust approach for patient-specific distortion correction. Initial benchmarking was conducted by scanning a phantom and generating B ₀ field maps with two approaches: (1) conventional B ₀ mapping and (2) experimental mDIXON imaging. Distortion maps were derived from the field maps and compared. The head and neck regions, including brain, of ten healthy volunteers were then evaluated at 1.5 T and 3 T. Distortion maps were again compared between approaches, using difference maps and histogram analysis. Overall, conventional B ₀ mapping was well approximated by mDIXON imaging: The distortions of 95% of the voxels in the phantom estimated by mDIXON and conventional B ₀ mapping differed by  <0.02 mm (1.5 T) and  <0.04 mm (3 T), while the 95-percentiles of the distortions estimated by conventional B ₀ mapping were  <0.06 mm (1.5 T) and  <0.12 mm (3 T). In head and neck the distortions of 99% of the voxels were within  ±0.2 mm at 1.5 T for both approaches and within  ±0.4 mm and  ±0.5 mm at 3 T for mDIXON imaging and conventional B ₀ mapping, respectively. The majority of differences in vivo were confined to regions with high spatial variation of the B ₀ field, mostly around internal air cavities. For 1.5 T, the mDIXON imaging-based correction alone reduced the 95-percentile of distortions from 0.15 mm to 0.03 mm and within the brain from 0.06 mm to 0.02 mm. Slightly lower reductions were observed at 3 T. In conclusion, mDIXON imaging closely approximated conventional B ₀ mapping for patient-specific distortion assessment. Estimates in the brain were in good agreement, and slight differences were observed near air/tissue interfaces in the head and neck. Overall, mDIXON imaging-based B ₀ field maps may be advantageous for rapid patient-specific distortion correction without additional imaging.

Automatic Pathological Lung Segmentation in Low-Dose CT Image Using Eigenspace Sparse Shape Composition

The segmentation of lungs with severe pathology is a nontrivial problem in the clinical application. Due to complex structures, pathological changes, individual differences, and low image quality, accurate lung segmentation in clinical 3-D computed tomography (CT) images is still a challenging task. To overcome these problems, a novel dictionary-based approach is introduced to automatically segment pathological lungs in 3-D low-dose CT images. Sparse shape composition is integrated with the eigenvector space shape prior model, called eigenspace sparse shape composition, to reduce local shape reconstruction error caused by the weak and misleading appearance prior information. To initialize the shape model, a landmark recognition method based on discriminative appearance dictionary is introduced to handle lesions and local details. Furthermore, a new vertex search strategy based on the gradient vector flow field is also proposed to drive the shape deformation to the target boundary. The proposed algorithm is tested on 78 3-D low-dose CT images with lung tumors. Compared to the state-of-the-art methods, the proposed approach can robustly and accurately detect pathological lung surface.

Machine learning for automated 3-dimensional segmentation of the spine and suggested placement of pedicle screws based on intraoperative cone-beam computer tomography

OBJECTIVE

The goal of this study was to develop and validate a system for automatic segmentation of the spine, pedicle identification, and screw path suggestion for use with an intraoperative 3D surgical navigation system.

METHODS

Cone-beam CT (CBCT) images of the spines of 21 cadavers were obtained. An automated model-based approach was used for segmentation. Using machine learning methodology, the algorithm was trained and validated on the image data sets. For measuring accuracy, surface area errors of the automatic segmentation were compared to the manually outlined reference surface on CBCT. To further test both technical and clinical accuracy, the algorithm was applied to a set of 20 clinical cases. The authors evaluated the system's accuracy in pedicle identification by measuring the distance between the user-defined midpoint of each pedicle and the automatically segmented midpoint. Finally, 2 independent surgeons performed a qualitative evaluation of the segmentation to judge whether it was adequate to guide surgical navigation and whether it would have resulted in a clinically acceptable pedicle screw placement.

RESULTS

The clinically relevant pedicle identification and automatic pedicle screw planning accuracy was 86.1%. By excluding patients with severe spinal deformities (i.e., Cobb angle > 75° and severe spinal degeneration) and previous surgeries, a success rate of 95.4% was achieved. The mean time (± SD) for automatic segmentation and screw planning in 5 vertebrae was 11 ± 4 seconds.

CONCLUSIONS

The technology investigated has the potential to aid surgeons in navigational planning and improve surgical navigation workflow while maintaining patient safety.

Fully automated prostate whole gland and central gland segmentation on MRI using holistically nested networks with short connections

Accurate and automated prostate whole gland and central gland segmentations on MR images are essential for aiding any prostate cancer diagnosis system. Our work presents a 2-D orthogonal deep learning method to automatically segment the whole prostate and central gland from T2-weighted axial-only MR images. The proposed method can generate high-density 3-D surfaces from low-resolution ( z axis) MR images. In the past, most methods have focused on axial images alone, e.g., 2-D based segmentation of the prostate from each 2-D slice. Those methods suffer the problems of over-segmenting or under-segmenting the prostate at apex and base, which adds a major contribution for errors. The proposed method leverages the orthogonal context to effectively reduce the apex and base segmentation ambiguities. It also overcomes jittering or stair-step surface artifacts when constructing a 3-D surface from 2-D segmentation or direct 3-D segmentation approaches, such as 3-D U-Net. The experimental results demonstrate that the proposed method achieves 92.4 % ± 3 % Dice similarity coefficient (DSC) for prostate and DSC of 90.1 % ± 4.6 % for central gland without trimming any ending contours at apex and base. The experiments illustrate the feasibility and robustness of the 2-D-based holistically nested networks with short connections method for MR prostate and central gland segmentation. The proposed method achieves segmentation results on par with the current literature.

Automatic multiorgan segmentation in thorax CT images using U-net-GAN

PURPOSE

Accurate and timely organs-at-risk (OARs) segmentation is key to efficient and high-quality radiation therapy planning. The purpose of this work is to develop a deep learning-based method to automatically segment multiple thoracic OARs on chest computed tomography (CT) for radiotherapy treatment planning.

METHODS

We propose an adversarial training strategy to train deep neural networks for the segmentation of multiple organs on thoracic CT images. The proposed design of adversarial networks, called U-Net-generative adversarial network (U-Net-GAN), jointly trains a set of U-Nets as generators and fully convolutional networks (FCNs) as discriminators. Specifically, the generator, composed of U-Net, produces an image segmentation map of multiple organs by an end-to-end mapping learned from CT image to multiorgan-segmented OARs. The discriminator, structured as an FCN, discriminates between the ground truth and segmented OARs produced by the generator. The generator and discriminator compete against each other in an adversarial learning process to produce the optimal segmentation map of multiple organs. Our segmentation results were compared with manually segmented OARs (ground truth) for quantitative evaluations in geometric difference, as well as dosimetric performance by investigating the dose-volume histogram in 20 stereotactic body radiation therapy (SBRT) lung plans.

RESULTS

This segmentation technique was applied to delineate the left and right lungs, spinal cord, esophagus, and heart using 35 patients' chest CTs. The averaged dice similarity coefficient for the above five OARs are 0.97, 0.97, 0.90, 0.75, and 0.87, respectively. The mean surface distance of the five OARs obtained with proposed method ranges between 0.4 and 1.5 mm on average among all 35 patients. The mean dose differences on the 20 SBRT lung plans are -0.001 to 0.155 Gy for the five OARs.

CONCLUSION

We have investigated a novel deep learning-based approach with a GAN strategy to segment multiple OARs in the thorax using chest CT images and demonstrated its feasibility and reliability. This is a potentially valuable method for improving the efficiency of chest radiotherapy treatment planning.

Integrated 3D Anatomical Model for Automatic Myocardial Segmentation in Cardiac CT Imagery

Segmentation of epicardial and endocardial boundaries is a critical step in diagnosing cardiovascular function in heart patients. The manual tracing of organ contours in Computed Tomography Angiography (CTA) slices is subjective, time-consuming and impractical in clinical setting. We propose a novel multi-dimensional automatic edge detection algorithm based on shape priors and principal component analysis (PCA). We have developed a highly customized parametric model for implicit representations of segmenting curves (3D) for Left Ventricle (LV), Right Ventricle (RV), and Epicardium (Epi) used simultaneously to achieve myocardial segmentation. We have combined these representations in a region-based image modeling framework with high level constraints enabling the modeling of complex cardiac anatomical structures to automatically guide the segmentation of endo/epicardial boundaries. Test results on 30 short-axis CTA datasets show robust segmentation with error (mean ± std mm) of (1.46 ± 0.41), (2.06 ± 0.65), (2.88 ± 0.59) for LV, RV and Epi respectively.

Organ-At-Risk Segmentation in Brain MRI using Model-Based Segmentation: Benefits of Deep Learning-Based Boundary Detectors

Organ-at-risk (OAR) segmentation is a key step for radiotherapy treatment planning. Model-based segmentation (MBS) has been successfully used for the fully automatic segmentation of anatomical structures and it has proven to be robust to noise due to its incorporated shape prior knowledge. In this work, we investigate the advantages of combining neural networks with the prior anatomical shape knowledge of the model-based segmentation of organs-at-risk for brain radiotherapy (RT) on Magnetic Resonance Imaging (MRI). We train our boundary detectors using two different approaches: classic strong gradients as described in [4] and as a locally adaptive regression task, where for each triangle a convolutional neural network (CNN) was trained to estimate the distances between the mesh triangles and organ boundary, which were then combined into a single network, as described by [1]. We evaluate both methods using a 5-fold cross- validation on both T1w and T2w brain MRI data from sixteen primary and metastatic brain cancer patients (some post-surgical). Using CNN-based boundary detectors improved the results for all structures in both T1w and T2w data. The improvements were statistically significant (p < 0.05) for all segmented structures in the T1w images and only for the auditory system in the T2w images.

Automated assessment of paraspinal muscle fat composition based on the segmentation of chemical shift encoding-based water/fat-separated images

Proton-density fat fraction (PDFF) of the paraspinal muscles, derived from chemical shift encoding-based water-fat magnetic resonance imaging, has emerged as an important surrogate biomarker in individuals with intervertebral disc disease, osteoporosis, sarcopenia and neuromuscular disorders. However, quantification of paraspinal muscle PDFF is currently limited in clinical routine due to the required time-consuming manual segmentation procedure. The present study aimed to develop an automatic segmentation algorithm of the lumbar paraspinal muscles based on water-fat sequences and compare the performance of this algorithm to ground truth data based on manual segmentation. The algorithm comprised an average shape model, a dual feature model, associating each surface point with a fat and water image appearance feature, and a detection model. Right and left psoas, quadratus lumborum and erector spinae muscles were automatically segmented. Dice coefficients averaged over all six muscle compartments amounted to 0.83 (range 0.75-0.90).

Registration-Based Organ Positioning and Joint Segmentation Method for Liver and Tumor Segmentation

The automated segmentation of liver and tumor from CT images is of great importance in medical diagnoses and clinical treatment. However, accurate and automatic segmentation of liver and tumor is generally complicated due to the complex anatomical structures and low contrast. This paper proposes a registration-based organ positioning (ROP) and joint segmentation method for liver and tumor segmentation from CT images. First, a ROP method is developed to obtain liver's bounding box accurately and efficiently. Second, a joint segmentation method based on fuzzy c-means (FCM) and extreme learning machine (ELM) is designed to perform coarse liver segmentation. Third, the coarse segmentation is regarded as the initial contour of active contour model (ACM) to refine liver boundary by considering the topological information. Finally, tumor segmentation is performed using another ELM. Experiments on two datasets demonstrate the performance advantages of our proposed method compared with other related works.

Dictionary-based electric properties tomography

PURPOSE

To develop and validate a new algorithm called "dictionary-based electric properties tomography" (dbEPT) for deriving tissue electric properties from measured B₁ maps.

METHODS

Inspired by Magnetic Resonance fingerprinting, dbEPT uses a dictionary of local patterns ("atoms") of B₁ maps and corresponding electric properties distributions, derived from electromagnetic field simulations. For reconstruction, a pattern from a measured B₁ map is compared with the B₁ atoms of the dictionary. The B₁ atom showing the best match with the measured B₁ pattern yields the optimum electric properties pattern that is chosen for reconstruction. Matching was performed through machine learning algorithms. Two dictionaries, using transmit and transceive phases, were evaluated. The spatial distribution of local matching distance between optimal atom and measured pattern yielded a reconstruction reliability map. The method was applied to reconstruct conductivity of 4 volunteers' brains. A conventional, Helmholtz-based Electric properties tomography (EPT) reconstruction was performed for reference. Noise performance was studied through phantom simulations.

RESULTS

Quantitative values of conductivity agree with literature values. Results of the 2 dictionaries exhibit only minor differences. Somewhat larger differences are visible between dbEPT and Helmholtz-based EPT. Quantified by the correlation between conductivity and anatomic images, dbEPT depicts brain details more clearly than Helmholtz-based EPT. Matching distance is minimal in homogeneous brain ventricles and increases with tissue heterogeneity. Central processing unit time was approximately 2 minutes per dictionary training and 3 minutes per brain conductivity reconstruction using standard hardware equipment.

CONCLUSION

A new, dictionary-based approach for reconstructing electric properties is presented. Its conductivity reconstruction is able to overcome the EPT transceive-phase problem.

SiSSR: Simultaneous subdivision surface registration for the quantification of cardiac function from computed tomography in canines

Recent improvements in cardiac computed tomography (CCT) allow for whole-heart functional studies to be acquired at low radiation dose (<2mSv) and high-temporal resolution (<100ms) in a single heart beat. Although the extraction of regional functional information from these images is of great clinical interest, there is a paucity of research into the quantification of regional function from CCT, contrasting with the large body of work in echocardiography and cardiac MR. Here we present the Simultaneous Subdivision Surface Registration (SiSSR) method: a fast, semi-automated image analysis pipeline for quantifying regional function from contrast-enhanced CCT. For each of thirteen adult male canines, we construct an anatomical reference mesh representing the left ventricular (LV) endocardium, obviating the need for a template mesh to be manually sculpted and initialized. We treat this generated mesh as a Loop subdivision surface, and adapt a technique previously described in the context of 3-D echocardiography to register these surfaces to the endocardium efficiently across all cardiac frames simultaneously. Although previous work performs the registration at a single resolution, we observe that subdivision surfaces naturally suggest a multiresolution approach, leading to faster convergence and avoiding local minima. We additionally make two notable changes to the cost function of the optimization, explicitly encouraging plausible biological motion and high mesh quality. Finally, we calculate an accepted functional metric for CCT from the registered surfaces, and compare our results to an alternate state-of-the-art CCT method.

Heart Chamber Segmentation from CT Using Convolutional Neural Networks

CT is routinely used for radiotherapy planning with organs and regions of interest being segmented for diagnostic evaluation and parameter optimization. For cardiac segmentation, many methods have been proposed for left ventricular segmentation, but few for simultaneous segmentation of the entire heart. In this work, we present a convolutional neural networks (CNN)-based cardiac chamber segmentation method for 3D CT with 5 classes: left ventricle, right ventricle, left atrium, right atrium, and background. We achieved an overall accuracy of 87.2% ± 3.3% and an overall chamber accuracy of 85.6 ± 6.1%. The deep learning based segmentation method may provide an automatic tool for cardiac segmentation on CT images.

Left Atrium Wall-mapping Application for Wall Thickness Visualisation

The measurement method for the LA wall thickness (WT) using cardiac computed tomography (CT) is observer dependent and cannot provide a rapid and comprehensive visualisation of the global LA WT. We aim to develop a LA wall-mapping application to display the global LA WT on a coplanar plane. The accuracy, intra-observer, and inter-observer reproducibility of the application were validated using digital/physical phantoms, and CT images of eight patients. This application on CT-based LA WT measures were further validated by testing six pig cardiac specimens. To evaluate its accuracy, the expanded maps of the physical phantom and pig LA were generated from the CT images and compared with the expanded map of the digital phantom and LA wall of pig heart. No significant differences (p > 0.05) were found between physical phantom and digital phantom as well as pig heart specimen and CT images using our application. Moreover, the analysis was based on the LA physical phantom or images of clinical patients; the results consistently demonstrated high intra-observer reproducibility (ICC > 0.9) and inter-observer reproducibility (ICC > 0.8) and showed good correlation between measures of pig heart specimen and CT data (r = 0.96, p < 0.001). The application can process and analyse the LA architecture for further visualisation and quantification.

Accuracy and precision of patient positioning for pelvic MR-only radiation therapy using digitally reconstructed radiographs

BACKGROUND AND PURPOSE

Magnetic resonance imaging (MRI) has in recent years emerged as an imaging modality to drive precise contouring of targets and organs at risk in external beam radiation therapy. Moreover, recent advances in MRI enable treatment of cancer without computed tomography (CT) simulation. A commercially available MR-only solution, MRCAT, offers a single-modality approach that provides density information for dose calculation and generation of positioning reference images. We evaluated the accuracy of patient positioning based on MRCAT digitally reconstructed radiographs (DRRs) by comparing to standard CT based workflow.

MATERIALS AND METHODS

Twenty consecutive prostate cancer patients being treated with external beam radiation therapy were included in the study. DRRs were generated for each patient based on the planning CT and MRCAT. The accuracy assessment was performed by manually registering the DRR images to planar kV setup images using bony landmarks. A Bayesian linear mixed effects model was used to separate systematic and random components (inter- and intra-observer variation) in the assessment. In addition, method agreement was assessed using a Bland-Altman analysis.

RESULTS

The systematic difference between MRCAT and CT based patient positioning, averaged over the study population, were found to be (mean [95% CI])  -0.49 [-0.85 to  -0.13] mm, 0.11 [-0.33 to  +0.57] mm and  -0.05 [-0.23 to  +0.36] mm in vertical, longitudinal and lateral directions, respectively. The increases in total random uncertainty were estimated to be below 0.5 mm for all directions, when using MR-only workflow instead of CT.

CONCLUSIONS

The MRCAT pseudo-CT method provides clinically acceptable accuracy and precision for patient positioning for pelvic radiation therapy based on planar DRR images. Furthermore, due to the reduction of geometric uncertainty, compared to dual-modality workflow, the approach is likely to improve the total geometric accuracy of pelvic radiation therapy.

A Meshfree Representation for Cardiac Medical Image Computing

The prominent advantage of meshfree method, is the way to build the representation of computational domain, based on the nodal points without any explicit meshing connectivity. Therefore, meshfree method can conveniently process the numerical computation inside interested domains with large deformation or inhomogeneity. In this paper, we adopt the idea of meshfree representation into cardiac medical image analysis in order to overcome the difficulties caused by large deformation and inhomogeneous materials of the heart. In our implementation, as element-free Galerkin method can efficiently build a meshfree representation using its shape function with moving least square fitting, we apply this meshfree method to handle large deformation or inhomogeneity for solving cardiac segmentation and motion tracking problems. We evaluate the performance of meshfree representation on a synthetic heart data and an in-vivo cardiac MRI image sequence. Results showed that the error of our framework against the ground truth was 0.1189 ± 0.0672 while the error of the traditional FEM was 0.1793 ± 0.1166. The proposed framework has minimal consistency constraints, handling large deformation and material discontinuities are simple and efficient, and it provides a way to avoid the complicated meshing procedures while preserving the accuracy with a relatively small number of nodes.

Very Small Left Atrial Volume as a Marker for Mortality in Patients Undergoing Nongated Computed Tomography Pulmonary Angiography

Objectives: To evaluate the association between very small left atria (VSLA) on nongated computed tomography pulmonary angiography (CTPA) and mortality in patients without pulmonary embolism (PE). Methods: Patients who underwent nongated CTPA between 2011 and 2015 in order to rule out PE, and had an echocardiogram within 24 h of the CTPA, were retrospectively identified. The left atrial volume of nongated CTPA was calculated using automatic 4-chamber volumetric analysis software. The association between the lowest 5th percentile of the left atrial volume index, referred to as the VSLA group, and mortality was investigated after adjustment for age, gender, background diseases, and laboratory values. Results: The study cohort included 241 patients. Patients with VSLA had a left atrial volume index <24 mL/m2 (n = 11). Demographics and background diseases did not differ between the study groups. The median follow-up was 22.7 months (IQR 0.03-54.3). VSLA was an independent predictor of mortality (HRadj = 3.6; 95% CI 1.46-8.87; p = 0.005), along with malignancy (HRadj = 2.28; 95% CI 1.32-3.93; p = 0.003) and lower hemoglobin (HRadj = 0.86; 95% CI 0.76-0.99; p = 0.032). Conclusions: Our findings suggest that VSLA on nongated CTPA may serve as a marker for mortality. The use of CTPA volumetric analysis can help risk stratification in patients with dyspnea and no PE.

Automated volumetric analysis of four cardiac chambers in pulmonary embolism

Identification of patients with acute pulmonary embolism (PE) who might be at risk of circulatory collapse by using a fast, automated system is highly desired. It was our objective to investigate whether automated cardiac volumetric analysis following computerised tomographic pulmonary angiography (CTPA) is useful to identify increased clot load and adverse prognosis in patients with acute PE. We retrospectively analysed a consecutive series of non-gated CTPA studies of 124 patients with acute PE and 43 controls. Right and left ventricular diameters (RV/LV) were measured on four-chamber view, while each cardiac chamber underwent automatic volumetric measurements. Findings were correlated to the pulmonary arterial obstruction index (PAOI). Outcome was expressed by admission to an intensive care unit (ICU) or mortality within 30 days. There was a significant positive correlation between the PAOI and the volumes of the right side cavities (r=0.25 for the atrium and r=0.49 for the ventricle), and between the right-to-left atrial and ventricular volume ratios (r=0.49 and r=0.57, respectively). Results for the combined outcome of mortality or ICU admission that fell in the upper tertile of the right atrial and right ventricular volumes yielded hazard ratios of 3.9 and 3.3, respectively, compared to those in the lower tertile. RV/LV diameter ratio did not correlate with outcome. In conclusion, adverse outcome and significant pulmonary clot load in patients with acute PE are associated with a volume shift towards right heart cavities, which correlates to prognosis better than the CT-measured RV/LV diameter ratio, suggesting the advantage of using fast fully automatic volumetric analysis to identify patients at risk.

A 3D Hermite-based multiscale local active contour method with elliptical shape constraints for segmentation of cardiac MR and CT volumes

Analysis of cardiac images is a fundamental task to diagnose heart problems. Left ventricle (LV) is one of the most important heart structures used for cardiac evaluation. In this work, we propose a novel 3D hierarchical multiscale segmentation method based on a local active contour (AC) model and the Hermite transform (HT) for LV analysis in cardiac magnetic resonance (MR) and computed tomography (CT) volumes in short axis view. Features such as directional edges, texture, and intensities are analyzed using the multiscale HT space. A local AC model is configured using the HT coefficients and geometrical constraints. The endocardial and epicardial boundaries are used for evaluation. Segmentation of the endocardium is controlled using elliptical shape constraints. The final endocardial shape is used to define the geometrical constraints for segmentation of the epicardium. We follow the assumption that epicardial and endocardial shapes are similar in volumes with short axis view. An initialization scheme based on a fuzzy C-means algorithm and mathematical morphology was designed. The algorithm performance was evaluated using cardiac MR and CT volumes in short axis view demonstrating the feasibility of the proposed method.

A competitive strategy for atrial and aortic tract segmentation based on deformable models

Multiple strategies have previously been described for atrial region (i.e. atrial bodies and aortic tract) segmentation. Although these techniques have proven their accuracy, inadequate results in the mid atrial walls are common, restricting their application for specific cardiac interventions. In this work, we introduce a novel competitive strategy to perform atrial region segmentation with correct delineation of the thin mid walls, and integrated it into the B-spline Explicit Active Surfaces framework. A double-stage segmentation process is used, which starts with a fast contour growing followed by a refinement stage with local descriptors. Independent functions are used to define each region, being afterward combined to compete for the optimal boundary. The competition locally constrains the surface evolution, prevents overlaps and allows refinement to the walls. Three different scenarios were used to demonstrate the advantages of the proposed approach, through the evaluation of its segmentation accuracy, and its performance for heterogeneous mid walls. Both computed tomography and magnetic resonance imaging datasets were used, presenting results similar to the state-of-the-art methods for both atria and aorta. The competitive strategy showed its superior performance with statistically significant differences against the traditional free-evolution approach in cases with bad image quality or missed atrial/aortic walls. Moreover, only the competitive approach was able to accurately segment the atrial/aortic wall. Overall, the proposed strategy showed to be suitable for atrial region segmentation with a correct segmentation of the mid thin walls, demonstrating its added value with respect to the traditional techniques.

[MRI-based radiotherapy planning]

MRI-based radiotherapy planning is a topical subject due to the introduction of a new generation of treatment machines combining a linear accelerator and a MRI. One of the issues for introducing MRI in this task is the lack of information to provide tissue density information required for dose calculation. To cope with this issue, two strategies may be distinguished from the literature. Either a synthetic CT scan is generated from the MRI to plan the dose, or a dose is generated from the MRI based on physical underpinnings. Within the first group, three approaches appear: bulk density mapping assign a homogeneous density to different volumes of interest manually defined on a patient MRI; machine learning-based approaches model local relationship between CT and MRI image intensities from multiple data, then applying the model to a new MRI; atlas-based approaches use a co-registered training data set (CT-MRI) which are registered to a new MRI to create a pseudo CT from spatial correspondences in a final fusion step. Within the second group, physics-based approaches aim at computing the dose directly from the hydrogen contained within the tissues, quantified by MRI. Excepting the physics approach, all these methods generate a synthetic CT called "pseudo CT", on which radiotherapy planning will be finally realized. This literature review shows that atlas- and machine learning-based approaches appear more accurate dosimetrically. Bulk density approaches are not appropriate for bone localization. The fastest methods are machine learning and the slowest are atlas-based approaches. The less automatized are bulk density assignation methods. The physical approaches appear very promising methods. Finally, the validation of these methods is crucial for a clinical practice, in particular in the perspective of adaptive radiotherapy delivered by a linear accelerator combined with an MRI scanner.

Identification of Pulmonary Hypertension Caused by Left-Sided Heart Disease (World Health Organization Group 2) Based on Cardiac Chamber Volumes Derived From Chest CT Imaging

BACKGROUND

Evaluations of patients with pulmonary hypertension (PH) commonly include chest CT imaging. We hypothesized that cardiac chamber volumes calculated from the same CT scans can yield additional information to distinguish PH related to left-sided heart disease (World Health Organization group 2) from other PH subtypes.

METHODS

Patients who had PH confirmed by right heart catheterization and contrast-enhanced chest CT studies were enrolled in this retrospective multicenter study. Cardiac chamber volumes were calculated using automated segmentation software and compared between group 2 and non-group 2 patients with PH.

RESULTS

This study included 114 patients with PH, 27 (24%) of whom were classified as group 2 based on their pulmonary capillary wedge pressure. Patients with group 2 PH exhibited significantly larger median left atrial (LA) volumes (118 mL vs 63 mL; P < .001), larger median left ventricular (LV) volumes (90 mL vs 76 mL; P = .02), and smaller median right ventricular (RV) volumes (173 mL vs 210 mL; P = .005) than did non-group 2 patients. On multivariate analysis adjusted for age, sex, and mean pulmonary arterial pressure, group 2 PH was significantly associated with larger median LA and LV volumes (P < .001 and P = .008, respectively) and decreased volume ratios of RA/LA, RV/LV, and RV/LA (P = .001, P = .004, and P < .001, respectively). Enlarged LA volumes demonstrated a high discriminatory ability for group 2 PH (area under the curve, 0.92; 95% CI, 0.870-0.968).

CONCLUSIONS

Volumetric analysis of the cardiac chambers from nongated chest CT scans, particularly with findings of an enlarged left atrium, exhibited high discriminatory ability for identifying patients with PH due to left-sided heart disease.

A model of cardiac contraction based on novel measurements of tension development in human cardiomyocytes

Experimental data from human cardiac myocytes at body temperature is crucial for a quantitative understanding of clinically relevant cardiac function and development of whole-organ computational models. However, such experimental data is currently very limited. Specifically, important measurements to characterize changes in tension development in human cardiomyocytes that occur with perturbations in cell length are not available. To address this deficiency, in this study we present an experimental data set collected from skinned human cardiac myocytes, including the passive and viscoelastic properties of isolated myocytes, the steady-state force calcium relationship at different sarcomere lengths, and changes in tension following a rapid increase or decrease in length, and after constant velocity shortening. This data set is, to our knowledge, the first characterization of length and velocity-dependence of tension generation in human skinned cardiac myocytes at body temperature. We use this data to develop a computational model of contraction and passive viscoelasticity in human myocytes. Our model includes troponin C kinetics, tropomyosin kinetics, a three-state crossbridge model that accounts for the distortion of crossbridges, and the cellular viscoelastic response. Each component is parametrized using our experimental data collected in human cardiomyocytes at body temperature. Furthermore we are able to confirm that properties of length-dependent activation at 37°C are similar to other species, with a shift in calcium sensitivity and increase in maximum tension. We revise our model of tension generation in the skinned isolated myocyte to replicate reported tension traces generated in intact muscle during isometric tension, to provide a model of human tension generation for multi-scale simulations. This process requires changes to calcium sensitivity, cooperativity, and crossbridge transition rates. We apply this model within multi-scale simulations of biventricular cardiac function and further refine the parametrization within the whole organ context, based on obtaining a healthy ejection fraction. This process reveals that crossbridge cycling rates differ between skinned myocytes and intact myocytes.

Automatic segmentation and quantification of the cardiac structures from non-contrast-enhanced cardiac CT scans

Early structural changes to the heart, including the chambers and the coronary arteries, provide important information on pre-clinical heart disease like cardiac failure. Currently, contrast-enhanced cardiac computed tomography angiography (CCTA) is the preferred modality for the visualization of the cardiac chambers and the coronaries. In clinical practice not every patient undergoes a CCTA scan; many patients receive only a non-contrast-enhanced calcium scoring CT scan (CTCS), which has less radiation dose and does not require the administration of contrast agent. Quantifying cardiac structures in such images is challenging, as they lack the contrast present in CCTA scans. Such quantification would however be relevant, as it enables population based studies with only a CTCS scan. The purpose of this work is therefore to investigate the feasibility of automatic segmentation and quantification of cardiac structures viz whole heart, left atrium, left ventricle, right atrium, right ventricle and aortic root from CTCS scans. A fully automatic multi-atlas-based segmentation approach is used to segment the cardiac structures. Results show that the segmentation overlap between the automatic method and that of the reference standard have a Dice similarity coefficient of 0.91 on average for the cardiac chambers. The mean surface-to-surface distance error over all the cardiac structures is [Formula: see text] mm. The automatically obtained cardiac chamber volumes using the CTCS scans have an excellent correlation when compared to the volumes in corresponding CCTA scans, a Pearson correlation coefficient (R) of 0.95 is obtained. Our fully automatic method enables large-scale assessment of cardiac structures on non-contrast-enhanced CT scans.

Comprehensive Modeling and Visualization of Cardiac Anatomy and Physiology from CT Imaging and Computer Simulations

In clinical cardiology, both anatomy and physiology are needed to diagnose cardiac pathologies. CT imaging and computer simulations provide valuable and complementary data for this purpose. However, it remains challenging to gain useful information from the large amount of high-dimensional diverse data. The current tools are not adequately integrated to visualize anatomic and physiologic data from a complete yet focused perspective. We introduce a new computer-aided diagnosis framework, which allows for comprehensive modeling and visualization of cardiac anatomy and physiology from CT imaging data and computer simulations, with a primary focus on ischemic heart disease. The following visual information is presented: (1) Anatomy from CT imaging: geometric modeling and visualization of cardiac anatomy, including four heart chambers, left and right ventricular outflow tracts, and coronary arteries; (2) Function from CT imaging: motion modeling, strain calculation, and visualization of four heart chambers; (3) Physiology from CT imaging: quantification and visualization of myocardial perfusion and contextual integration with coronary artery anatomy; (4) Physiology from computer simulation: computation and visualization of hemodynamics (e.g., coronary blood velocity, pressure, shear stress, and fluid forces on the vessel wall). Substantially, feedback from cardiologists have confirmed the practical utility of integrating these features for the purpose of computer-aided diagnosis of ischemic heart disease.

A Novel Mouse Segmentation Method Based on Dynamic Contrast Enhanced Micro-CT Images

With the development of hybrid imaging scanners, micro-CT is widely used in locating abnormalities, studying drug metabolism, and providing structural priors to aid image reconstruction in functional imaging. Due to the low contrast of soft tissues, segmentation of soft tissue organs from mouse micro-CT images is a challenging problem. In this paper, we propose a mouse segmentation scheme based on dynamic contrast enhanced micro-CT images. With a homemade fast scanning micro-CT scanner, dynamic contrast enhanced images were acquired before and after injection of non-ionic iodinated contrast agents (iohexol). Then the feature vector of each voxel was extracted from the signal intensities at different time points. Based on these features, the heart, liver, spleen, lung, and kidney could be classified into different categories and extracted from separate categories by morphological processing. The bone structure was segmented using a thresholding method. Our method was validated on seven BALB/c mice using two different classifiers: a support vector machine classifier with a radial basis function kernel and a random forest classifier. The results were compared to manual segmentation, and the performance was assessed using the Dice similarity coefficient, false positive ratio, and false negative ratio. The results showed high accuracy with the Dice similarity coefficient ranging from 0.709 ± 0.078 for the spleen to 0.929 ± 0.006 for the kidney.

Using physiologically based models for clinical translation: predictive modelling, data interpretation or something in-between?

Heart disease continues to be a significant clinical problem in Western society. Predictive models and simulations that integrate physiological understanding with patient information derived from clinical data have huge potential to contribute to improving our understanding of both the progression and treatment of heart disease. In particular they provide the potential to improve patient selection and optimisation of cardiovascular interventions across a range of pathologies. Currently a significant proportion of this potential is still to be realised. In this paper we discuss the opportunities and challenges associated with this realisation. Reviewing the successful elements of model translation for biophysically based models and the emerging supporting technologies, we propose three distinct modes of clinical translation. Finally we outline the challenges ahead that will be fundamental to overcome if the ultimate goal of fully personalised clinical cardiac care is to be achieved.

Automatic Four-Chamber Segmentation Using Level-Set Method and Split Energy Function

Objectives

In this paper, we present an automatic method to segment four chambers by extracting a whole heart, separating the left and right sides of the heart, and spliting the atrium and ventricle regions from each heart in cardiac computed tomography angiography (CTA) efficiently.

Methods

We smooth the images by applying filters to remove noise. Next, the volume of interest is detected by using k-means clustering. In this step, the whole heart is coarsely extracted, and it is used for seed volumes in the next step. Then, we detect seed volumes using a geometric analysis based on anatomical information and separate the left and right heart regions with the power watershed algorithm. Finally, we refine the left and right sides of the heart using the level-set method, and extract the atrium and ventricle from the left and right heart regions using the split energy function.

Results

We tested the proposed heart segmentation method using 20 clinical scan datasets which were acquired from various patients. To validate the proposed heart segmentation method, we evaluated its accuracy in segmenting four chambers based on four error evaluation metrics. The average values of differences between the manual and automatic segmentations were less than 3.3%, approximately.

Conclusions

The proposed method extracts the four chambers of the heart accurately, demonstrating that this approach can assist the cardiologist.

Fast automated segmentation of multiple objects via spatially weighted shape learning

Active shape models (ASMs) have proved successful in automatic segmentation by using shape and appearance priors in a number of areas such as prostate segmentation, where accurate contouring is important in treatment planning for prostate cancer. The ASM approach however, is heavily reliant on a good initialisation for achieving high segmentation quality. This initialisation often requires algorithms with high computational complexity, such as three dimensional (3D) image registration. In this work, we present a fast, self-initialised ASM approach that simultaneously fits multiple objects hierarchically controlled by spatially weighted shape learning. Prominent objects are targeted initially and spatial weights are progressively adjusted so that the next (more difficult, less visible) object is simultaneously initialised using a series of weighted shape models. The scheme was validated and compared to a multi-atlas approach on 3D magnetic resonance (MR) images of 38 cancer patients and had the same (mean, median, inter-rater) Dice's similarity coefficients of (0.79, 0.81, 0.85), while having no registration error and a computational time of 12-15 min, nearly an order of magnitude faster than the multi-atlas approach.

Fully Automatic 3-D-TEE Segmentation for the Planning of Transcatheter Aortic Valve Implantation

A novel fully automatic framework for aortic valve (AV) trunk segmentation in three-dimensional (3-D) transesophageal echocardiography (TEE) datasets is proposed. The methodology combines a previously presented semiautomatic segmentation strategy by using shape-based B-spline Explicit Active Surfaces with two novel algorithms to automate the quantification of relevant AV measures. The first combines a fast rotation-invariant 3-D generalized Hough transform with a vessel-like dark tube detector to initialize the segmentation. After segmenting the AV wall, the second algorithm focuses on aligning this surface with the reference ones in order to estimate the short-axis (SAx) planes (at the left ventricular outflow tract, annulus, sinuses of Valsalva, and sinotubular junction) in which to perform the measurements. The framework has been tested in 20 3-D-TEE datasets with both stenotic and nonstenotic AVs. The initialization algorithm presented a median error of around 3 mm for the AV axis endpoints, with an overall feasibility of 90%. In its turn, the SAx detection algorithm showed to be highly reproducible, with indistinguishable results compared with the variability found between the experts' defined planes. Automatically extracted measures at the four levels showed a good agreement with the experts' ones, with limits of agreement similar to the interobserver variability. Moreover, a validation set of 20 additional stenotic AV datasets corroborated the method's applicability and accuracy. The proposed approach mitigates the variability associated with the manual quantification while significantly reducing the required analysis time (12 s versus 5 to 10 min), which shows its appeal for automatic dimensioning of the AV morphology in 3-D-TEE for the planning of transcatheter AV implantation.

Automatic multiorgan segmentation in CT images of the male pelvis using region-specific hierarchical appearance cluster models

PURPOSE

Accurate segmentation of pelvic organs in CT images is of great importance in external beam radiotherapy for prostate cancer. The aim of this studying is to develop a novel method for automatic, multiorgan segmentation of the male pelvis.

METHODS

The authors' segmentation method consists of several stages. First, a pretreatment includes parameterization, principal component analysis (PCA), and an established process of region-specific hierarchical appearance cluster (RSHAC) model which was executed on the training dataset. After the preprocessing, online automatic segmentation of new CT images is achieved by combining the RSHAC model with the PCA-based point distribution model. Fifty pelvic CT from eight prostate cancer patients were used as the training dataset. From another 20 prostate cancer patients, 210 CT images were used for independent validation of the segmentation method.

RESULTS

In the training dataset, 15 PCA modes were needed to represent 95% of shape variations of pelvic organs. When tested on the validation dataset, the authors' segmentation method had an average Dice similarity coefficient and mean absolute distance of 0.751 and 0.371 cm, 0.783 and 0.303 cm, 0.573 and 0.604 cm for prostate, bladder, and rectum, respectively. The automated segmentation process took on average 5 min on a personal computer equipped with Core 2 Duo CPU of 2.8 GHz and 8 GB RAM.

CONCLUSIONS

The authors have developed an efficient and reliable method for automatic segmentation of multiple organs in the male pelvis. This method should be useful for treatment planning and adaptive replanning for prostate cancer radiotherapy. With this method, the physicist can improve the work efficiency and stability.

ECG imaging of ventricular tachycardia: evaluation against simultaneous non-contact mapping and CMR-derived grey zone

ECG imaging is an emerging technology for the reconstruction of cardiac electric activity from non-invasively measured body surface potential maps. In this case report, we present the first evaluation of transmurally imaged activation times against endocardially reconstructed isochrones for a case of sustained monomorphic ventricular tachycardia (VT). Computer models of the thorax and whole heart were produced from MR images. A recently published approach was applied to facilitate electrode localization in the catheter laboratory, which allows for the acquisition of body surface potential maps while performing non-contact mapping for the reconstruction of local activation times. ECG imaging was then realized using Tikhonov regularization with spatio-temporal smoothing as proposed by Huiskamp and Greensite and further with the spline-based approach by Erem et al. Activation times were computed from transmurally reconstructed transmembrane voltages. The results showed good qualitative agreement between the non-invasively and invasively reconstructed activation times. Also, low amplitudes in the imaged transmembrane voltages were found to correlate with volumes of scar and grey zone in delayed gadolinium enhancement cardiac MR. The study underlines the ability of ECG imaging to produce activation times of ventricular electric activity-and to represent effects of scar tissue in the imaged transmembrane voltages.

Unsupervised Freeview Groupwise Cardiac Segmentation Using Synchronized Spectral Network

The diagnosis, comparative and population study of cardiac radiology data require heart segmentation on increasingly large amount of images from different modalities/chambers/patients under various imaging views. Most existing automatic cardiac segmentation methods are often limited to single image segmentation with regulated modality/region settings or well-cropped ROI areas, which is impossible for large datasets due to enormous device protocols and institutional differences. A pure data-driven unsupervised segmentation without regulated setting requirements is crucial in this scenario, and will significantly automate the manual work and adopt the various changes of modality, subject or view. In this paper, we propose a general unsupervised groupwise segmentation: a direct simultaneous segmentation for a group of multi-modality, multi-chamber, multi-subject ( M³) cardiac images from a freely chosen imaging view. The segmentation can directly perform not only on regulated two/four-chamber images, but also on non-regulated uncropped raw MR/CT scans. A new Synchronized Spectral Network (SSN) is developed for the simultaneous decomposing, synchronizing, and clustering the spectral features of free-view M³ cardiac images. The SSN-based groupwise analysis of image spectral bases immediately leads to groupwise segmentation of M³ freeview images. The segmentation is quantitatively evaluated by three datasets (MR and CT mixed) with more than 200 subjects. High dice metric ( ) is consistently achieved in validation. Our method provides a powerful tool for medical images under general imaging environment.

Standardized Evaluation System for Left Ventricular Segmentation Algorithms in 3D Echocardiography

Real-time 3D Echocardiography (RT3DE) has been proven to be an accurate tool for left ventricular (LV) volume assessment. However, identification of the LV endocardium remains a challenging task, mainly because of the low tissue/blood contrast of the images combined with typical artifacts. Several semi and fully automatic algorithms have been proposed for segmenting the endocardium in RT3DE data in order to extract relevant clinical indices, but a systematic and fair comparison between such methods has so far been impossible due to the lack of a publicly available common database. Here, we introduce a standardized evaluation framework to reliably evaluate and compare the performance of the algorithms developed to segment the LV border in RT3DE. A database consisting of 45 multivendor cardiac ultrasound recordings acquired at different centers with corresponding reference measurements from three experts are made available. The algorithms from nine research groups were quantitatively evaluated and compared using the proposed online platform. The results showed that the best methods produce promising results with respect to the experts' measurements for the extraction of clinical indices, and that they offer good segmentation precision in terms of mean distance error in the context of the experts' variability range. The platform remains open for new submissions.

Automatic bone removal for 3D TACE planning with C-arm CBCT: Evaluation of technical feasibility

PURPOSE

To evaluate the technical feasibility of automatically removing the ribs and spine from C-arm cone-beam computed tomography (CBCT) images acquired during transcatheter arterial chemoembolization (TACE).

MATERIAL AND METHODS

Fifty-eight patients (45.8 ± 5.0 years) with unresectable hepatocellular carcinoma (HCC) underwent transcatheter arterial chemoembolization and had intraprocedural CBCT imaging. Automatic bone removal was performed using model-based segmentation of the ventral cavity. Two interventional radiologists independently evaluated the performance of bone removal, remaining soft tissue retention, and general usability (where both the bone is appropriately removed while retaining soft tissue) for 3D TACE planning on a four-level (complete/excellent, adequate/good, incomplete/questionable, insufficient/bad) score. The proportion of inter-reader agreement was calculated.

RESULTS

For ribs and spine removal, 98.3-100% and 100% of cases showed complete or adequate performance, respectively. In 96.6% of the cases, soft tissue was at least adequately retained. 91.3-93.1% of the cases demonstrated good or excellent general usability for TACE planning. Satisfactory inter-reader agreement proportion was achieved in ribs (93.1%) and spine removal (89.7%), soft tissue retention (84.5%), and general usability for TACE planning (72.4%).

CONCLUSION

Intraprocedural automatic bone removal on CBCT images is technically feasible and offers good removal of ribs and spine while preserving soft tissue. Its clinical value needs further assessment.