Atlas-Based Ventricular Shape Analysis for Understanding Congenital Heart Disease

A deep learning approach for fully automated cardiac shape modeling in tetralogy of Fallot

BACKGROUND

Cardiac shape modeling is a useful computational tool that has provided quantitative insights into the mechanisms underlying dysfunction in heart disease. The manual input and time required to make cardiac shape models, however, limits their clinical utility. Here we present an end-to-end pipeline that uses deep learning for automated view classification, slice selection, phase selection, anatomical landmark localization, and myocardial image segmentation for the automated generation of three-dimensional, biventricular shape models. With this approach, we aim to make cardiac shape modeling a more robust and broadly applicable tool that has processing times consistent with clinical workflows.

METHODS

Cardiovascular magnetic resonance (CMR) images from a cohort of 123 patients with repaired tetralogy of Fallot (rTOF) from two internal sites were used to train and validate each step in the automated pipeline. The complete automated pipeline was tested using CMR images from a cohort of 12 rTOF patients from an internal site and 18 rTOF patients from an external site. Manually and automatically generated shape models from the test set were compared using Euclidean projection distances, global ventricular measurements, and atlas-based shape mode scores.

RESULTS

The mean absolute error (MAE) between manually and automatically generated shape models in the test set was similar to the voxel resolution of the original CMR images for end-diastolic models (MAE = 1.9 ± 0.5 mm) and end-systolic models (MAE = 2.1 ± 0.7 mm). Global ventricular measurements computed from automated models were in good agreement with those computed from manual models. The average mean absolute difference in shape mode Z-score between manually and automatically generated models was 0.5 standard deviations for the first 20 modes of a reference statistical shape atlas.

CONCLUSIONS

Using deep learning, accurate three-dimensional, biventricular shape models can be reliably created. This fully automated end-to-end approach dramatically reduces the manual input required to create shape models, thereby enabling the rapid analysis of large-scale datasets and the potential to deploy statistical atlas-based analyses in point-of-care clinical settings. Training data and networks are available from cardiacatlas.org.

An Atlas-Based Analysis of Biventricular Mechanics in Tetralogy of Fallot

The current study proposes an efficient strategy for exploiting the statistical power of cardiac atlases to investigate whether clinically significant variations in ventricular shape are sufficient to explain corresponding differences in ventricular wall motion directly, or if they are indirect markers of altered myocardial mechanical properties. This study was conducted in a cohort of patients with repaired tetralogy of Fallot (rTOF) that face long-term right ventricular (RV) and/or left ventricular (LV) dysfunction as a consequence of adverse remodeling. Features of biventricular end-diastolic (ED) shape associated with RV apical dilation, LV dilation, RV basal bulging, and LV conicity correlated with components of systolic wall motion (SWM) that contribute most to differences in global systolic function. A finite element analysis of systolic biventricular mechanics was employed to assess the effect of perturbations in these ED shape modes on corresponding components of SWM. Perturbations to ED shape modes and myocardial contractility explained observed variation in SWM to varying degrees. In some cases, shape markers were partial determinants of systolic function and, in other cases, they were indirect markers for altered myocardial mechanical properties. Patients with rTOF may benefit from an atlas-based analysis of biventricular mechanics to improve prognosis and gain mechanistic insight into underlying myocardial pathophysiology.

A Population-Based 3D Atlas of the Pathological Lumbar Spine Segment

The spine is the load-bearing structure of human beings and may present several disorders, with low back pain the most frequent problem during human life. Signs of a spine disorder or disease vary depending on the location and type of the spine condition. Therefore, we aim to develop a probabilistic atlas of the lumbar spine segment using statistical shape modeling (SSM) and then explore the variability of spine geometry using principal component analysis (PCA). Using computed tomography (CT), the human spine was reconstructed for 24 patients with spine disorders and then the mean shape was deformed upon specific boundaries (e.g., by ±3 or ±1.5 standard deviation). Results demonstrated that principal shape modes are associated with specific morphological features of the spine segment such as Cobb’s angle, lordosis degree, spine width and height. The lumbar spine atlas here developed has evinced the potential of SSM to investigate the association between shape and morphological parameters, with the goal of developing new treatments for the management of patients with spine disorders.

Right Ventricular Flow Vorticity Relationships With Biventricular Shape in Adult Tetralogy of Fallot

Remodeling in adults with repaired tetralogy of Fallot (rToF) may occur due to chronic pulmonary regurgitation, but may also be related to altered flow patterns, including vortices. We aimed to correlate and quantify relationships between vorticity and ventricular shape derived from atlas-based analysis of biventricular shape. Adult rToF (n = 12) patients underwent 4D flow and cine MRI imaging. Vorticity in the RV was computed after noise reduction using a neural network. A biventricular shape atlas built from 95 rToF patients was used to derive principal component modes, which were associated with vorticity and pulmonary regurgitant volume (PRV) using univariate and multivariate linear regression. Univariate analysis showed that indexed PRV correlated with 3 modes (r = −0.55,−0.50, and 0.6, all p < 0.05) associated with RV dilatation and an increase in basal bulging, apical bulging and tricuspid annulus tilting with more severe regurgitation, as well as a smaller LV and paradoxical movement of the septum. RV outflow and inflow vorticity were also correlated with these modes. However, total vorticity over the whole RV was correlated with two different modes (r = −0.62,−0.69, both p < 0.05). Higher vorticity was associated with both RV and LV shape changes including longer ventricular length, a larger bulge beside the tricuspid valve, and distinct tricuspid tilting. RV flow vorticity was associated with changes in biventricular geometry, distinct from associations with PRV. Flow vorticity may provide additional mechanistic information in rToF remodeling. Both LV and RV shapes are important in rToF RV flow patterns.

Statistical shape modeling of multi-organ anatomies with shared boundaries

Introduction: Statistical shape modeling (SSM) is a valuable and powerful tool to generate a detailed representation of complex anatomy that enables quantitative analysis of shapes and their variations. SSM applies mathematics, statistics, and computing to parse the shape into some quantitative representation (such as correspondence points or landmarks) which can be used to study the covariance patterns of the shapes and answer various questions about the anatomical variations across the population. Complex anatomical structures have many diverse parts with varying interactions or intricate architecture. For example, the heart is a four-chambered organ with several shared boundaries between chambers. Subtle shape changes within the shared boundaries of the heart can indicate potential pathologic changes such as right ventricular overload. Early detection and robust quantification could provide insight into ideal treatment techniques and intervention timing. However, existing SSM methods do not explicitly handle shared boundaries which aid in a better understanding of the anatomy of interest. If shared boundaries are not explicitly modeled, it restricts the capability of the shape model to identify the pathological shape changes occurring at the shared boundary. Hence, this paper presents a general and flexible data-driven approach for building statistical shape models of multi-organ anatomies with shared boundaries that explicitly model contact surfaces. Methods: This work focuses on particle-based shape modeling (PSM), a state-of-art SSM approach for building shape models by optimizing the position of correspondence particles. The proposed PSM strategy for handling shared boundaries entails (a) detecting and extracting the shared boundary surface and contour (outline of the surface mesh/isoline) of the meshes of the two organs, (b) followed by a formulation for a correspondence-based optimization algorithm to build a multi-organ anatomy statistical shape model that captures morphological and alignment changes of individual organs and their shared boundary surfaces throughout the population. Results: We demonstrate the shared boundary pipeline using a toy dataset of parameterized shapes and a clinical dataset of the biventricular heart models. The shared boundary model for the cardiac biventricular data achieves consistent parameterization of the shared surface (interventricular septum) and identifies the curvature of the interventricular septum as pathological shape differences.

All Roads Lead to Rome: Diverse Etiologies of Tricuspid Regurgitation Create a Predictable Constellation of Right Ventricular Shape Changes

Introduction: Myriad disorders cause right ventricular (RV) dilation and lead to tricuspid regurgitation (TR). Because the thin-walled, flexible RV is mechanically coupled to the pulmonary circulation and the left ventricular septum, it distorts with any disturbance in the cardiopulmonary system. TR, therefore, can result from pulmonary hypertension, left heart failure, or intrinsic RV dysfunction; but once it occurs, TR initiates a cycle of worsening RV volume overload, potentially progressing to right heart failure. Characteristic three-dimensional RV shape-changes from this process, and changes particular to individual TR causes, have not been defined in detail. Methods: Cardiac MRI was obtained in 6 healthy volunteers, 41 patients with ≥ moderate TR, and 31 control patients with cardiac disease without TR. The mean shape of each group was constructed using a three-dimensional statistical shape model via the particle-based shape modeling approach. Changes in shape were examined across pulmonary hypertension and congestive heart failure subgroups using principal component analysis (PCA). A logistic regression approach based on these PCA modes identified patients with TR using RV shape alone. Results: Mean RV shape in patients with TR exhibited free wall bulging, narrowing of the base, and blunting of the RV apex compared to controls (p < 0.05). Using four primary PCA modes, a logistic regression algorithm identified patients with TR correctly with 82% recall and 87% precision. In patients with pulmonary hypertension without TR, RV shape was narrower and more streamlined than in healthy volunteers. However, in RVs with TR and pulmonary hypertension, overall RV shape continued to demonstrate the free wall bulging characteristic of TR. In the subgroup of patients with congestive heart failure without TR, this intermediate state of RV muscular hypertrophy was not present. Conclusion: The multiple causes of TR examined in this study changed RV shape in similar ways. Logistic regression classification based on these shape changes reliably identified patients with TR regardless of etiology. Furthermore, pulmonary hypertension without TR had unique shape features, described here as the "well compensated" RV. These results suggest shape modeling as a promising tool for defining severity of RV disease and risk of decompensation, particularly in patients with pulmonary hypertension.

Computational analysis of cardiac structure and function in congenital heart disease: Translating discoveries to clinical strategies

Increased availability and access to medical image data has enabled more quantitative approaches to clinical diagnosis, prognosis, and treatment planning for congenital heart disease. Here we present an overview of long-term clinical management of tetralogy of Fallot (TOF) and its intersection with novel computational and data science approaches to discovering biomarkers of functional and prognostic importance. Efforts in translational medicine that seek to address the clinical challenges associated with cardiovascular diseases using personalized and precision-based approaches are then discussed. The considerations and challenges of translational cardiovascular medicine are reviewed, and examples of digital platforms with collaborative, cloud-based, and scalable design are provided.

Personalising left-ventricular biophysical models of the heart using parametric physics-informed neural networks

We present a parametric physics-informed neural network for the simulation of personalised left-ventricular biomechanics. The neural network is constrained to the biophysical problem in two ways: (i) the network output is restricted to a subspace built from radial basis functions capturing characteristic deformations of left ventricles and (ii) the cost function used for training is the energy potential functional specifically tailored for hyperelastic, anisotropic, nearly-incompressible active materials. The radial bases are generated from the results of a nonlinear Finite Element model coupled with an anatomical shape model derived from high-resolution cardiac images. We show that, by coupling the neural network with a simplified circulation model, we can efficiently generate computationally inexpensive estimations of cardiac mechanics. Our model is 30 times faster than the reference Finite Element model used, including training time, while yielding satisfactory average errors in the predictions of ejection fraction (-3%), peak systolic pressure (7%), stroke work (4%) and myocardial strains (14%). This physics-informed neural network is well suited to efficiently augment cardiac images with functional data and to generate large sets of synthetic cases for training deep network classifiers while it provides efficient personalization to the specific patient of interest with a high level of detail.

Right Ventricular Shape Distortion in Tricuspid Regurgitation

Tricuspid regurgitation (TR) is a failure in right-sided AV valve function which, if left untreated, leads to marked cardiac shape changes and heart failure. However, the specific right ventricular shape changes resulting from TR are unknown. The goal of this study is to characterize the RV shape changes of patients with severe TR. RVs were segmented from CINE MRI images. Using particle-based shape modeling (PSM), a dense set of homologous landmarks were placed with geometric consistency on the endocardial surface of each RV, via an entropy-based optimization of the information content of the shape model. Principal component analysis (PCA) identified the significant modes of shape variation across the population. These modes were used to create a patient-prediction model. 32 patients and 6 healthy controls were studied. The mean RV shape of TR patients demonstrated increased sphericity relative to controls, with the three most dominant modes of variation showing significant widening of the short axis of the heart, narrowing of the base at the RV outflow tract (RVOT), and blunting of the RV apex. By PCA, shape changes based on the first three modes of variation correctly identified patient vs. control hearts 86.5% of the time. The shape variation may further illuminate the mechanics of TR-induced RV failure and recovery, providing potential targets for therapies including novel devices and surgical interventions.

Atlas-based measures of left ventricular shape may improve characterization of adverse remodeling in anthracycline-exposed childhood cancer survivors: a cross-sectional imaging study

Background

Adverse cardiac remodeling is an important precursor to anthracycline-related cardiac dysfunction, however conventional remodeling indices are limited. We sought to examine the utility of statistical atlas-derived measures of ventricular shape to improve the identification of adverse anthracycline-related remodeling in childhood cancer survivors.

Methods

We analyzed cardiac magnetic resonance imaging from a cross-sectional cohort of 20 childhood cancer survivors who were treated with low (< 250 mg/m² [N = 10]) or high (≥250 mg/m² [N = 10]) dose anthracyclines, matched 1:1 by sex and age between dose groups. We reconstructed 3D computational models of left ventricular end-diastolic shape for each subject and assessed the ability of conventional remodeling indices (volume, mass, and mass to volume ratio) vs. shape modes derived from a statistical shape atlas of an asymptomatic reference population to stratify anthracycline-related remodeling. We compared conventional parameters and five atlas-based shape modes: 1) between survivors and the reference population (N = 1991) using multivariable linear regression, and 2) within survivors by anthracycline dose (low versus high) using two-sided T-tests, multivariable logistic regression, and receiver operating characteristic curves.

Results

Compared with the reference population, survivors had differences in conventional measures (lower volume and mass) and shape modes (corresponding to lower overall size and lower sphericity; all p < 0.001). Among survivors, differences in a shape mode corresponding to increased basal cavity size and altered mitral annular orientation in the high-dose group were observed (p = 0.039). Collectively, atlas-based shape modes in conjunction with conventional measures discriminated survivors who received low vs. high anthracycline dosage (area under the curve [AUC] 0.930, 95% confidence interval 0.816, 1.00) significantly better than conventional measures alone (AUC 0.710, 95% confidence interval 0.473, 0.947; AUC comparison p = 0.0498).

Conclusions

Compared with a reference population, heart size is smaller in anthracycline-exposed childhood cancer survivors. Atlas-based measures of left ventricular shape may improve the detection of anthracycline dose-related remodeling differences.

Artificial Intelligence in Cardiac Imaging With Statistical Atlases of Cardiac Anatomy

In many cardiovascular pathologies, the shape and motion of the heart provide important clues to understanding the mechanisms of the disease and how it progresses over time. With the advent of large-scale cardiac data, statistical modeling of cardiac anatomy has become a powerful tool to provide automated, precise quantification of the status of patient-specific heart geometry with respect to reference populations. Powered by supervised or unsupervised machine learning algorithms, statistical cardiac shape analysis can be used to automatically identify and quantify the severity of heart diseases, to provide morphometric indices that are optimally associated with clinical factors, and to evaluate the likelihood of adverse outcomes. Recently, statistical cardiac atlases have been integrated with deep neural networks to enable anatomical consistency of cardiac segmentation, registration, and automated quality control. These combinations have already shown significant improvements in performance and avoid gross anatomical errors that could make the results unusable. This current trend is expected to grow in the near future. Here, we aim to provide a mini review highlighting recent advances in statistical atlasing of cardiac function in the context of artificial intelligence in cardiac imaging.

Combining Statistical Shape Model and Principal Component Analysis to Estimate Left Ventricular Volume and Ejection Fraction

Left ventricular ejection fraction (LVEF) assessment is instrumental for cardiac health diagnosis, patient management, and patient eligibility for participation in clinical studies. Due to its non-invasiveness and low operational cost, ultrasound (US) imaging is the most commonly used imaging modality to image the heart and assess LVEF. Even though 3D US imaging technology is becoming more available, cardiologists dominantly use 2D US imaging to visualize the LV blood pool and interpret its area changes between end-systole and end-diastole. Our previous work showed that LVEF estimates based on area changes are significantly lower than the true volume-based estimates by as much as 13%,¹ which could lead to unnecessary and costly therapeutic decisions. Acquiring volumetric information about the LV blood pool necessitates either time-consuming 3D reconstruction or 3D US image acquisition. Here, we propose a method that leverages on a statistical shape model (SSM) constructed from 13 landmarks depicting the LV endocardial border to estimate a new patient's LV volume and LVEF. Two methods to estimate the 3D LV geometry with and without size normalization were employed. The SSM was built using the 13 landmarks from 50 training patient image datasets. Subsequently, the Mahalanobis distance (with size normalization) or the vector distance (without size normalization) between an incoming patient's LV landmarks and each shape in the SSM were used to determine the weights each training patient contributed to describing the new, incoming patient's LV geometry and associated blood pool volume. We tested the proposed method to estimate the LV volumes and LVEF for 16 new test patients. The estimated LVEFs based on Mahalanobis distance and vector distance were within 2.9% and 1.1%, respectively, of the ground truth LVEFs calculated from the 3D reconstructed LV volumes. Furthermore, the viability of using fewer principal components (PCs) to estimate the LV volume was explored by reducing the number of PCs retained when projecting landmarks onto PCA space. LVEF estimated based on 3 PCs, 5 PCs, and 10 PCs are within 6.6%, 5.4%, and 3.3%, respectively, of LVEF estimates using the full set of 39 PCs.

A Statistical Shape Model Approach for Computing Left Ventricle Volume and Ejection Fraction Using Multi-plane Ultrasound Images

Assessing the left ventricular ejection fraction (LVEF) accurately requires 3D volumetric data of the LV. Cardiologists either have no access to 3D ultrasound (US) systems or prefer to visually estimate LVEF based on 2D US images. To facilitate the consistent estimation of the end-diastolic and end-systolic blood pool volume and LVEF based on 3D data without extensive complicated manual input, we propose a statistical shape model (SSM) based on 13 key anchor points-the LV apex (1), mitral valve hinges (6), and the midpoints of the endocardial contours (6)-identified from the LV endocardial contour of the tri-plane 2D US images. We use principal component analysis (PCA) to identify the principle modes of variation needed to represent the LV shapes, which enables us to estimate an incoming LV as a linear combination of the principle components (PC). For a new, incoming patient image, its 13 anchor points are projected onto the PC space; its shape is compared to each LV shape in the SSM based on Mahalanobis distance, which is normalized with respect to the LV size, as well as direct vector distance (i.e., PCA distance), without any size normalization. These distances are used to determine the weight each training shape in the SSM contributes to the description of the new patient LV shape. Finally, the new patient's LV systolic and diastolic volumes are estimated as the weighted average of the training volumes in the SSM. To assess our proposed method, we compared the SSM-based estimates of diastolic, systolic, stroke volumes, and LVEF with those computed directly from 16 tri-plane 2D US imaging datasets using the GE Echo-Pac PC clinical platform. The estimated LVEF based on Mahalanobis distance and PCA distance were within 6.8% and 1.7% of the reference LVEF computed using the GE Echo-Pac PC clinical platform.

Modeling left ventricular dynamics with characteristic deformation modes

A computationally efficient method is described for simulating the dynamics of the left ventricle (LV) in three dimensions. LV motion is represented as a combination of a limited number of deformation modes, chosen to represent observed cardiac motions while conserving volume in the LV wall. The contribution of each mode to wall motion is determined by a corresponding time-dependent deformation variable. The principle of virtual work is applied to these deformation variables, yielding a system of ordinary differential equations for LV dynamics, including effects of muscle fiber orientations, active and passive stresses, and surface tractions. Passive stress is governed by a transversely isotropic elastic model. Active stress acts in the fiber direction and incorporates length-tension and force-velocity properties of cardiac muscle. Preload and afterload are represented by lumped vascular models. The variational equations and their numerical solutions are verified by comparison to analytic solutions of the strong form equations. Deformation modes are constructed using Fourier series with an arbitrary number of terms. Greater numbers of deformation modes increase deformable model resolution but at increased computational cost. Simulations of normal LV motion throughout the cardiac cycle are presented using models with 8, 23, or 46 deformation modes. Aggregate quantities that describe LV function vary little as the number of deformation modes is increased. Spatial distributions of stress and strain change as more deformation modes are included, but overall patterns are conserved. This approach yields three-dimensional simulations of the cardiac cycle on a clinically relevant time-scale.

Statistical Shape Modeling for Cavopulmonary Assist Device Development: Variability of Vascular Graft Geometry and Implications for Hemodynamics

Patients born with a single functional ventricle typically undergo three-staged surgical palliation in the first years of life, with the last stage realizing a cross-like total cavopulmonary connection (TCPC) of superior and inferior vena cavas (SVC and IVC) with both left and right pulmonary arteries, allowing all deoxygenated blood to flow passively back to the lungs (Fontan circulation). Even though within the past decades more patients survive into adulthood, the connection comes at the prize of deficiencies such as chronic systemic venous hypertension and low cardiac output, which ultimately may lead to Fontan failure. Many studies have suggested that the TCPC's inherent insufficiencies might be addressed by adding a cavopulmonary assist device (CPAD) to provide the necessary pressure boost. While many device concepts are being explored, few take into account the complex cardiac anatomy typically associated with TCPCs. In this study, we focus on the extra cardiac conduit vascular graft connecting IVC and pulmonary arteries as one possible landing zone for a CPAD and describe its geometric variability in a cohort of 18 patients that had their TCPC realized with a 20mm vascular graft. We report traditional morphometric parameters and apply statistical shape modeling to determine the main contributors of graft shape variability. Such information may prove useful when designing CPADs that are adapted to the challenging anatomical boundaries in Fontan patients. We further compute the anatomical mean 3D graft shape (template graft) as a representative of key shape features of our cohort and prove this template graft to be a significantly better approximation of population and individual patient's hemodynamics than a commonly used simplified tube geometry. We therefore conclude that statistical shape modeling results can provide better models of geometric and hemodynamic boundary conditions associated with complex cardiac anatomy, which in turn may impact on improved cardiac device development.

Investigating Cardiac Motion Patterns Using Synthetic High-Resolution 3D Cardiovascular Magnetic Resonance Images and Statistical Shape Analysis

Diagnosis of ventricular dysfunction in congenital heart disease is more and more based on medical imaging, which allows investigation of abnormal cardiac morphology and correlated abnormal function. Although analysis of 2D images represents the clinical standard, novel tools performing automatic processing of 3D images are becoming available, providing more detailed and comprehensive information than simple 2D morphometry. Among these, statistical shape analysis (SSA) allows a consistent and quantitative description of a population of complex shapes, as a way to detect novel biomarkers, ultimately improving diagnosis and pathology understanding. The aim of this study is to describe the implementation of a SSA method for the investigation of 3D left ventricular shape and motion patterns and to test it on a small sample of 4 congenital repaired aortic stenosis patients and 4 age-matched healthy volunteers to demonstrate its potential. The advantage of this method is the capability of analyzing subject-specific motion patterns separately from the individual morphology, visually and quantitatively, as a way to identify functional abnormalities related to both dynamics and shape. Specifically, we combined 3D, high-resolution whole heart data with 2D, temporal information provided by cine cardiovascular magnetic resonance images, and we used an SSA approach to analyze 3D motion per se. Preliminary results of this pilot study showed that using this method, some differences in end-diastolic and end-systolic ventricular shapes could be captured, but it was not possible to clearly separate the two cohorts based on shape information alone. However, further analyses on ventricular motion allowed to qualitatively identify differences between the two populations. Moreover, by describing shape and motion with a small number of principal components, this method offers a fully automated process to obtain visually intuitive and numerical information on cardiac shape and motion, which could be, once validated on a larger sample size, easily integrated into the clinical workflow. To conclude, in this preliminary work, we have implemented state-of-the-art automatic segmentation and SSA methods, and we have shown how they could improve our understanding of ventricular kinetics by visually and potentially quantitatively highlighting aspects that are usually not picked up by traditional approaches.