Statistical Atlases and Computational Models of the Heart. Imaging and Modelling Challenges

In order to assess the variability in the calculation of the pressure gradient through a moderate thoracic aortic coarctation (MTAC), a 3D finite element model of MTAC was constructed, which includes the ascending aorta, the aortic arch, the descending aorta, and the three large branches (the innominate artery, the left common carotid artery, and the left subclavian artery), as well as with a coarctation in the descending aorta. The surface model of MTAC in STL format was imported into ANSYS ICEM CFD12.1 to generate volume mesh.A finite element model suitable for hemodynamics analysis of patient-specific MTAC was established.Numerical simulation of hemodynamics in this model was performed by means of Computational fluid dynamics (CFD) using ANSYS CFX12.1. The temporal distributions of homodynamic variables such as streamlines, wall pressure, velocity vector and wall shear stress in the arteries were analyzed during a cardiac cycle. The maximum and the average of pressure gradient in a cardiac cycle through a MTAC are 13 mmHg and 2.84 mmHg respectively. The pressure difference between the systolic and the diastolic in a cardiac cycle proximal to the coarctation is about 38 mmHg, which is smaller than the difference between the recorded systolic and diastolic pressures of 115 and 65 mmHg (i.e. the difference is 115-65=50). Similarly, the pressure gradient through the coarctation under exercise conditions could be predicted via modifying the inflow and outflow boundary conditions under resting conditions. CFD techniques make it possible to obtain information (such as pressure when the patient is under exercise condition) which is difficult to get in clinic practice or in experiment based on patient-specific data.

Pressure gradient across coarctation of aorta (CoA) is conventionally computed from phase contrast magnetic resonance imaging (PC-MRI) by applying the Bernoulli equation to the peak blood flow velocity measurement obtained just distal to the aortic narrowing. In order to test the validity and accuracy of the Bernoulli flow assumptions of negligible viscous forces in assessment of pressure gradients across the coarctation, we sought to determine pressure information from patient-specific computational fluid dynamics (CFD) simulation, modeling Newtonian, viscous, incompressible blood flow under steady and pulsatile inflow conditions. The transient high velocity jet observed though a moderate thoracic aortic coarctation model (65% area reduction) reconstructed from magnetic resonance angiography scans of an 8-year old female patient provided for the 2012 STACOM CFD challenge, was studied over a cardiac cycle under patient-specific flow conditions. Descending aorta hemodynamics was contrasted with a geometrically and dynamically comparable normal aorta simulation. The peak velocity of the modeled CoA jet (6.99 m/s) was observed to occur ~2 cm distal to the site of coarctation. The magnitude of this velocity was found to be similar to appropriately dynamically scaled clinical observations (6.00±0.6 m/s) of peak velocity obtained from PC-MRI data on three pre-surgical CoA patients, evaluated at Children’s Hospital of Pittsburgh. Bernoulli pressure gradient across the CoA computed using the CFD velocity field at the peak-systole instant of pulsatile flow grossly overestimated the true gradient predicted from CFD (30 mm Hg) when unsteady jet wake effects were more pronounced, but underestimated the CFD pressure gradient at steady time-averaged inflow conditions (5.8 mm Hg). Based on this pilot study, CFD determined flow fields are a more reliable clinical indicator of pressure gradient which considers viscous flow and complex jet wake interactions affecting hemodynamics downstream of CoA.

In this paper, we propose a system to determine the pressure gradient at rest in the aorta. We developed a technique to efficiently initialize a regular simulation grid from a patient-specific aortic triangulated model. On this grid we employ the lattice Boltzmann method to resolve the characteristic fluid flow through the vessel. The inflow rates, as measured physiologically, are imposed providing accurate pulsatile flow. The simulation required a resolution of at least 20 microns to ensure a convergence of the pressure calculation. HARVEY, a large-scale parallel code, was run on the IBM Blue Gene/Q supercomputer to model the flow at this high resolution. We analyze and evaluate the strengths and weaknesses of our system.

Aortic Coarctation is a congenital constriction of the aorta that increases blood pressure above the constriction and hinders the flow below it. Based on a 3D surface mesh of a moderate thoracic coarctation, a high quality volume mesh is created using an optimal tetrahedral aspect ratio for whole domain. In order to quantify the severity of this constriction, a coupled 1D lumped-parameter/3D CFD approach is used to calculate the pressure drop through the coarctation. The CFD computation is performed assuming that the arterial wall is rigid and the blood is considered a homogeneous Newtonian fluid with density r = 0.001 gr/mm3 and a dynamic viscosity m = 0.004 gr/mm/sec in laminar flow. The boundary conditions of the 3D model (inlet and outlet conditions) have been calculated using a 1D model. Parallelization procedures will be used in order to increase the performance of the CFD calculations.

We investigate a patient specific blood flow simulation through a transverse aortic arch with a moderate thoracic aortic coarctation, where particular attention is paid to the blood pressure gradient through the coarctation. The challenge in this context is the complex geometry containing a stenosis, which results in complex flow patterns. The fluid is assumed to be incompressible and Newtonian. Its dynamic is usually described by an Navier-Stokes equation with appropriate boundary conditions. Instead, we modeled the problem mesoscopically by a family of BGK-Boltzmann equations those solutions reaches that of a corresponding Navier-Stokes system in a certain limit. For discretization we take advantage of lattice Boltzmann methods, which are realized within the open-source library OpenLB. A realistic transient flow profile of the cardiac output for a human at rest was used to specify the inflow boundary condition at the aortic root, whereas the outflow at the descending aorta was modeled by a pressure boundary condition. A short introduction to lattice Boltzmann methods is provided and especially the used boundary conditions are introduced in detail. The exact simulation setup is stated and the obtained results are discussed.

This work presents our approach for modelling the CFD challenge example of the aortic coarctation of an 8 year old child. The three-dimensional fluid domain was modeled as described in the challenge as an incompressible Newtonian fluid. A residual-based variational multiscale finite element method is used to solve the 3D fluid field. The boundaries were treated with 3-element windkessel models. The windkessel elements were tuned using an adjoint based method to fit the pressure and flowrate values reported by the challenge. A mesh refinement was performend to ensure the spatial convergence of the presented results. Finally, pressure values at

π

1

and

π

2

slices are reported.

Delayed Enhancement Magnetic Resonance Imaging can be used to non-invasively differentiate viable from non-viable myocardium within the Left Ventricle in patients suffering from myocardial diseases. Automated segmentation of scarified tissue can be used to accurately quantify the percentage of myocardium affected. This paper presents a method for cardiac scar detection and segmentation based on supervised learning and level set segmentation. First, a model of the appearance of scar tissue is trained using a Support Vector Machines classifier on image-derived descriptors. Based on the areas detected by the classifier, an accurate segmentation is performed using a segmentation method based on level sets.

We propose an automatic technique to segment scar and classify the myocardial tissue of the left ventricle from Delay Enhancement (DE) MRI. The method uses multiple region growing with two types of regions and automatic seed initialization. The region growing criteria is based on intensity distance and the seed initialization is based on a thresholding technique. We refine the obtained segmentation with some morphological operators and geometrical constraints to further define the infarcted area. Thanks to the use of two types of regions when performing the region growing, we are able to segment and classify the healthy and pathological tissues. We have also a third type of tissue in our classification, which includes tissue areas that deserve special attention from medical experts: border-zone tissue or myocardial segmentation errors.

Delayed-enhancement magnetic resonance imaging (DE-MRI) is an effective technique for imaging left ventricular (LV) infarct. Existing techniques for LV infarct segmentation are primarily threshold-based making them prone to high user variability. In this work, we propose a segmentation algorithm that can learn from training images and segment based on this training model. This is implemented as a Markov random field (MRF) based energy formulation solved using graph-cuts. A good agreement was found with the Full-Width-at-Half-Maximum (FWHM) technique.

Accurate detection and delineation of myocardium infarction is important for treatment planning in patients with heart disease. Delayed contrast enhanced magnetic resonance imaging (DE-MRI) is a well established technique for the assessment of myocardial infarction. However, manual delineation of myocardium infarction in DE-MRI is both time consuming and prone to intra and inter rater variability. In this paper, we present an automatic, probabilistic framework for segmentation of myocardium infarction using Hierarchical Conditional Random Fields (HCRFs). In each level, a CRF classifier with up to triplet clique potentials is learnt. Furthermore, incorporation of spin image features in the second level allows for better learning the neighbourhood characteristics. The performance of the HCRF classifier on 5 animal scans and 5 human scans shows promising results.

Myocardial viability assessment is an important task in the diagnosis of coronary heart disease. The measurement of the delayed enhancement effect, the accumulation of contrast agent in defective tissue, has become the gold standard for detecting necrotic tissue with MRI. The purpose of the presented work was to provide a segmentation and quantification method for delayed enhancement MRI. To this end, a suitable mixture model for the myocardial intensity distribution is determined based on expectation maximization and the comparison of the fit accuracy. The subsequent watershed-based segmentation uses the intensity threshold information derived from this model. Preliminary results are derived from an analysis of datasets provided by the STACOM challenge organizers. The segmentation provided reasonable results in all datasets, but the method strongly depends on the underlying myocardium segmentation.

This paper presents collated results from the Delayed Enhancement MRI (DE-MRI) segmentation challenge at MICCAI 2012. DE-MRI Images from fifteen patients and fifteen pigs were randomly selected from two different imaging centres. Three independent sets of manual segmentations were obtained for each image and included in this study. A ground truth consensus segmentation based on all human rater segmentations was obtained using an Expectation-Maximization (EM) method (the STAPLE method). Automated segmentations from five groups contributed to this challenge.

Cardiac magnetic resonance imaging (MRI) is a key diagnostic tool for non-invasive assessment of the function and structure of the cardiovascular system in clinical practice. Cardiac landmarks provide strong cues to navigate the complex heart anatomy, extract and evaluate morphological and functional features for diagnosis and disease monitoring. A fully automatic method is presented to detect cardiac landmarks from individual images using a learning-based approach to model discriminative context. In addition to the target landmarks, auxiliary markers are taken into consideration to construct context with more discriminative power. The presented approach is evaluated on the STACOM2012 database, containing 100 independent test cases. Automatic landmark detection targets include two mitral valve landmarks in a long axis image, two RV insert landmarks in a short-axis image, and one central axis point in an LV base image.

We propose a supervised learning approach for detecting landmarks in cardiac images from different views. A set of candidate landmark points are obtained using morphological operations and graph cut segmentation. The final landmarks are determined using random forests (RF) classifiers which were trained on low level features derived from the neighborhood of annotated landmarks on training images. We use features like intensity, texture, shape asymmetry and context information for landmark detection. Experimental results on the STACOM LV landmark detection challenge dataset show that our approaching is promising with room for further improvement.

This paper describes the data setup of the second cardiac Motion Analysis Challenge (cMac2). The purpose of this challenge is to initiate a public data repository for the benchmark of motion and strain quantification algorithms on 3D ultrasound images. The data currently includes synthetic images that combine ultrasound and biomechanical simulators. We also collected sonomicrometry curves and ultrasound images acquired on a Polyvinyl alcohol phantom.

This paper presents motion and deformation quantification results obtained from

synthetic

and

in vitro

phantom data provided by the second

cardiac Motion Analysis Challenge

at STACOM-MICCAI. We applied the Temporal Diffeomorphic Free Form Deformation (TDFFD) algorithm to the datasets. This algorithm builds upon a diffeomorphic version of the FFD, to provide a 3

D

 + 

t

continuous and differentiable transform. The similarity metric includes a comparison between consecutive images, and between a reference and each of the following images.

Motion and strain accuracy were evaluated on synthetic 3D ultrasound sequences with known ground truth motion. Experiments were also conducted on

in vitro

acquisitions.

We present a novel method for tracking myocardial motion from volumetric ultrasound data based on non-rigid image registration using an anatomical free-form deformation model. Traditionally, the B-spline control points of such a model are defined on a rectangular grid in Cartesian space. This arrangement may be suboptimal as it treats the blood pool and myocardium similarly and as it enforces spatial smoothness in non-physiological directions. In this work, the basis functions are locally oriented along the radial, longitudinal and circumferential direction of the endocardium. This formulation allows us to model the left ventricular motion more naturally. We obtained encouraging accuracy results for the simulated models, with average errors of 0.8±0.6mm (10% relatively) and 0.5±0.4mm (15% relatively) compared to the ground truth in high and low contractility models respectively.

This paper describes an algorithm for motion and deformation quantification of 3D cardiac ultrasound sequences. The algorithm is based on the assumption that the deformation field is smooth inside the myocardium. Thus, we assume that the displacement field can be represented as the convolution of an unknown field with a Gaussian kernel. We apply our algorithm to datasets with reliable ground truth: a set of synthetic sequences with known trajectories and a set of sequences of a mechanical phantom implanted with microsonometry crystals.

We present a method for the analysis of heart motion from 3D cardiac ultrasound sequences. The algorithm exploits the monogenic signal theory, recently introduced as a N-dimensional generalization of the analytic signal. The displacement is computed locally by tracking variations in the monogenic phase. A 3D local affine displacement model accounts for typical motions as contraction/expansion and shearing. A coarse-to-fine B-spline scheme allows a robust and effective computation of the model parameters and a pyramidal refinement scheme helps in dealing with large motions. The independence of the monogenic phase on the local energy makes the algorithm insensitive to the time variant changes of image intensity that are often observed on echocardiographic sequences. The performance of our method is evaluated on 10 realistic simulated 3D echocardiographic sequences, showing good tracking accuracy (average error: 0.68±0.5 to 1.27±0.9 mm).

Analysis of echocardiograms is a valuable tool for assessing myocardial function and diseases. Processing of ultrasound data is challenging due to noise levels and depth-dependent quality of structure edges. We propose to adapt a method based on quadrature filters that is invariant to changes in intensity and has been successfully applied to MRI data earlier. Quadrature-filter-based registration derives the spatial deformation between two images from the local phase shift. Because the local phase is intensity-invariant and requires inhomogeneity, e.g., noise and intensity variations, to properly pick up phase shifts, it is well suited for ultrasound data. A multi-resolution and multi-scale scheme is used to cover different scales of deformations. The type and strength of regularization of the dense deformation field can be specified for each level, allowing for weighting of global and local motion. To speed up the registration, deformation fields are determined slice-wise for three orientations of the original data and subsequently combined into a true 3D deformation field. The method is evaluated with the data and ground truth provided by the

Cardiac Motion Analysis Challenge

at STACOM 2012.

In this paper, we evaluate the iLogDemons algorithm for the STACOM 2012 cardiac motion tracking challenge. This algorithm was previously applied to the STACOM 2011 cardiac motion challenge to track the left-ventricle heart tissue in a data-set of volunteers. Even though the previous application showed reasonable results with respect to quality of the registration and computed strain curves; quantitative evaluation of the algorithm in an objective manner is still not trivial. Applying the algorithm to the STACOM 2012 synthetic ultrasound sequence helps to objectively evaluate the algorithm since the ground truth motion is provided. Different configurations of the iLogDemons parameters are used and the estimated left ventricle motion is compared to the ground truth motion. Using this application, quantitative measurements of the motion error are calculated and optimal parameters of the algorithm can be found.

Regional wall motion and infarct scoring of MR images are routine clinical tools to grade performance and scarring in the heart. The aim of this paper is to provide a framework for automatic scoring to alert the diagnostician to potential regions of abnormality. We investigated different shape and motion configurations of a finite-element cardiac atlas of the left ventricle. Two patient populations were used: 300 asymptomatic volunteers and 105 patients with myocardial infarction, both randomly selected from the Cardiac Atlas Project database. Support vector machines were employed to estimate the boundaries between the asymptomatic control and patient groups for each of 16 standard anatomical regions in the heart. Ground truth visual wall motion scores from standard cines and infarct scoring from late enhancement were provided by experienced observers. From all configurations, end-systolic shape best predicted wall motion abnormalities (global accuracy 78%, positive predictive value 85%, specificity 91%, sensitivity 60%) and infarct scoring (74%, 72%, 91%, 44%). In conclusion, computer assisted wall motion and infarct scoring has the potential to provide robust identification of those segments requiring further clinical attention; in particular, the high specificity and relatively low sensitivity could help avoid unnecessary late gadolinium rescanning of patients.

X-ray fluoroscopic images are widely used for image guidance in cardiac electrophysiology (EP) procedures to diagnose or treat cardiac arrhythmias based on catheter ablation. However, the main disadvantage of fluoroscopic imaging is the lack of soft tissue information and harmful radiation. In contrast, ultrasound (US) has the advantages of low-cost, non-radiation, and high contrast in soft tissue. In this paper we propose a framework to extract the catheter from both X-ray and US images in real time for cardiac interventions. The catheter extraction from X-ray images is based on SURF features, local patch analysis and Kalman filtering to acquire a set of sorted key points representing the catheter. At the same time, the transformation between the X-ray and US images can be obtained via 2D/3D rigid registration between a 3D model of the US probe and its projection on X-ray images. By backprojecting the information about the catheter location in the X-ray images to the US images the search space can be drastically reduced. The extraction of the catheter from US is based on 3D SURF feature clusters, graph model building, A* algorithm and B-spline smoothing. Experiments show the overall process can be achieved in 2.72 seconds for one frame and the reprojected error is 1.99 mm on average.

The ventricular myocardium has a structure of branching laminae through which course regularly orientated fibers, an architecture important in excitation and contraction. DT-MRI is used to measure the fiber and laminar orientations. We quantify the performance of DT-MRI and structure tensor (ST) analysis of 3D high resolution MRI in five rat hearts and validate these against manual measurements. The ST and DT data are more similar for measures of the fiber orientation than laminar orientation. The average angle differences of elevation angles are 2.3±27.2˚, R = 0.57 for the fiber, 3.62±36.2˚, R = 0.24 for the laminae and 10.7±37.9˚, R = 0.32 for the laminae normal. The difference between DT and manually measured laminar orientation is 17±15˚ for DT and 5±10˚ for ST. DT and ST are comparable measures of the fiber orientation but ST is a better measure of myolaminar orientation.

Noninvasive blood flow measurements by 4D flow-sensitive MRI can be used to compute the intravascular distribution of blood pressure. In this work, we present an efficient algorithm for this task, based on the Navier-Stokes equations including zero-divergence condition for the velocity field. Its accuracy and robustness is investigated on two different CFD-based software phantoms. The method has been integrated into research software for analysis of clinical 4D flow measurements, and is tested on six patients with aortic coarctation. The pressure drop across the stenosis is quantified and coincides well with published results from a previously validated solution algorithm.

X-ray angiography is the most common imaging modality employed in the diagnosis of coronary diseases prior to or during a catheter-based intervention. The analysis of the patient X-Ray sequence can provide useful information about the degree of arterial stenosis, the myocardial perfusion and other clinical parameters. If the sequence has been acquired to evaluate the perfusion grade, the opacity due to the diaphragm could potentially hinder any kind of visual inspection and make more difficult a computer aided measurements. In this paper we propose an accurate and robust method to automatically identify the diaphragm border in each frame. Quantitative evaluation on a set of 11 sequences shows that the proposed algorithm outperforms previous methods.

Many cardiovascular diseases are linked to anomalies in myocardial fibers. The purpose of this paper is to model the birefringence of myocardial fibers in polarized light imaging (PLI) with future application to measurements on real myocardial tissues. The method consists in modeling the behavior of a uni-axial birefringent crystal by means of the Muller matrix, and measuring the final intensity of polarized light and consequently the orientation of myocardial fibers, by using crossed polarizers. The method was illustrated with a tissue modeled as a volume of 100×100×500

μ

m

3

. This volume was divided into cubes of size 20

μ

m close to cell diameter. The fiber orientation within the cube was defined by azimuth and elevation angles. The results showed that the proposed modeling enables us to find the optimal conditions for the PLI of 3D fiber orientations and design a model for the myocardial tissue measurement from PLI.

We propose an angle independent method for quantification of flow through composite surfaces covering the cross section of cardiac inflow and outflow tracts. We interleave trigger-delayed 3D colour Doppler sequences to increase frame rate and use multiple views to increase coverage.

Our method is applied to four patients with Hypoplastic Left Heart Syndrome (HLHS). Flow and velocity measurement are compared to Phase Contrast Magnetic Resonance Imaging (PC-MRI). Results are highly time-resolved and agree well with PC-MRI and show superior performance compared to standard measurements. Mean dissimilarity with respect to PC-MRI was found to be 7.35% ±3.72% (neoaortic outflow) and 10.15% ±2.72% (tricuspid inflow).

Computational simulations of the heart are a powerful tool for a comprehensive understanding of cardiac function and its intrinsic relationship with its muscular architecture. Cardiac biomechanical models require a vector field representing the orientation of cardiac fibers. A wrong orientation of the fibers can lead to a non-realistic simulation of the heart functionality.

In this paper we explore the impact of the fiber information on the simulated biomechanics of cardiac muscular anatomy. We have used the John Hopkins database to perform a biomechanical simulation using both a synthetic benchmark fiber distribution and the data obtained experimentally from DTI. Results illustrate how differences in fiber orientation affect heart deformation along cardiac cycle.

Step criterion edge detector (STEP) has been employed for the detection of endocardial edges in a Kalman filter based left ventricle tracking framework in previous studies. STEP determines the endocardial edge positions by fitting piecewise constant functions to intensity profiles, which are extracted on a tracked surface’s normal directions. In this study, we generalize STEP using higher order piecewise polynomial functions. The generalized STEP detectors make different assumptions about the endocardial edge representations, and their accuracies vary over the endocardial surface and cardiac cycle positions. Accordingly, we combine the responses of the generalized detectors using a maximum likelihood (ML) approach. Unlike previously proposed ML approaches, our combined edge detector provides a real-time tracking solution as the majority of regressive functions for the polynomial fitting can be computed offline. Comparative analyses showed that the combined detector (1) outperforms each of the generalized STEP detectors, and (2) provides a comparable accuracy with the previously defined slower ML approach.

Even using objective measures from DT-MRI no consensus about myocardial architecture has been achieved so far. Streamlining provides good reconstructions at low level of detail, but falls short to give global abstract interpretations. In this paper, we present a multi-resolution methodology that is able to produce simplified representations of cardiac architecture. Our approach produces a reduced set of tracts that are representative of the main geometric features of myocardial anatomical structure. Experiments show that fiber geometry is preserved along reductions, which validates the simplified model for interpretation of cardiac architecture.

Whole organ scale patient specific biophysical simulations contribute to the understanding, diagnosis and treatment of complex diseases such as cardiac arrhythmia. However, many individual steps are required to bridge the gap from an anatomical scan to a personalized biophysical model. In biophysical modeling, differential equations are solved on spatial domains represented by volumetric meshes of high resolution and in model-based segmentation, surface or volume meshes represent the patient’s geometry. We simplify the personalization process by representing the simulation mesh and additional relevant structures relative to the segmentation mesh. Using a surface correspondence preserving model-based segmentation algorithm, we facilitate the integration of anatomical information into biophysical models avoiding a complex processing pipeline. In a simulation study, we observe surface correspondence of up to 1.6 mm accuracy for the four heart chambers. We compare isotropic and anisotropic atrial excitation propagation in a personalized simulation.

Understanding the motion of the heart through the cardiac cycle can give useful insight for a range of different pathologies. In particular, quantifying regional cardiac motion can help clinicians to better determine cardiac function by identifying regions of thickened, ischemic or infarcted tissue. In this work we propose a method for cardiac motion analysis to track the deformation of the left ventricle at a regional level. This method estimates the affine motion of distinct regions of the myocardium using a near incompressible non-rigid registration algorithm based on the Demon’s optical flow approach. The global motion over the ventricle is computed by a smooth fusion of the deformation in each segment using an anatomically aware poly-affine model for the heart. We apply the proposed method to a data-set of 10 volunteers. The results indicate that we are able to extract reasonably realistic deformation fields parametrised by a significantly reduced number of parameters compared to voxel-wise methods, which better enables for statistical analyses of the motion.

This work aims at developing a training simulator for interventional radiology and thermo-ablation of cardiac arrhythmias. To achieve this, a real-time model of the cardiac electrophysiology is needed, which is very challenging due to the stiff equations involved. In this paper, we detail our contributions in order to obtain efficient cardiac electrophysiology simulations. First, an adaptive parametrisation of the Mitchell-Schaeffer model as well as numerical optimizations are proposed. An accurate computation of both conduction velocity and action potential is ensured, even with relatively coarse meshes. Second, a GPU implementation of the electrophysiology was realised in order to decrease the computation time. We evaluate our results by comparison with an accurate reference simulation using model parameters, personalized on patient data. We demonstrate that a fast simulation (close to real-time) can be obtained while keeping a precise description of the phenomena.

We have developed finite element modelling techniques to semi-automatically generate personalised biomechanical models of the human left ventricle (LV) based on cardiac magnetic resonance images. Geometric information of the LV throughout the cardiac cycle was derived via semi-automatic segmentation using non-rigid image registration with a pre-segmented image. A reference finite element mechanics model was automatically fitted to the segmented LV endocardial and epicardial surface data at diastasis. Passive and contractile myocardial mechanical properties were then tuned to best match the segmented surface data at end-diastole and end-systole, respectively. Global and regional indices of myocardial mechanics, including muscle fibre stress and extension ratio were then quantified and analysed. This mechanics modelling framework was applied to a healthy human subject and a patient with non-ischaemic heart failure. Comparison of the estimated passive stiffness and maximum activation level between the normal and diseased cases provided some preliminary insight into the changes in myocardial mechanical properties during heart failure. This automated approach enables minimally invasive personalised characterisation of cardiac mechanical function in health and disease. It also has the potential to elucidate the mechanisms of heart failure, and provide new quantitative diagnostic markers and therapeutic strategies for heart failure.

For X-ray guided cardiac catheterisations it is desirable to overlay pre-operative 3D-CT information for additional guidance. A common technique to obtain an overlay is to loop a catheter inside a target chamber and then manually align CT-derived surface models using multiple X-ray views. We propose and test a fully automatic algorithm for this purpose. The algorithm aligns the images by first estimating the pose of the CT relative to the X-ray table using the supine and isocentre constraints. Subsequently the pose is refined using an iterative optimisation strategy that maximises the intersection between the loop and projected target chamber, while minimising the separation between the X-ray cardiac border and the projected ventricles, in two X-ray views. Validation was carried out using a geometrically-realistic plastic heart phantom with two looped-catheter configurations formed inside the left atrium. The algorithm executes in under five minutes and yields average 3D-TREs between 2.4 and 5.4 mm over various regions of the heart, and 4.0 mm over the four chambers. Preliminary evaluation of this novel approach indicates feasibility for clinical interventional guidance and merits thorough validation using further phantom and clinical images.

The purpose of this work was to use

in vivo

MR imaging and electro-anatomical maps to characterize dense scars and border zone, BZ (a mixture of collagen and viable fibers). To better understand how these measures might probe potentially arrhythmogenic substrates, we developed a preclinical swine model of chronic infarction and integrated

in vivo

MRI and electrophysiology (EP) data in five swine at 5-6 weeks post-infarction. Specifically, we first aligned and registered T

1

-maps (from MR studies) and bipolar voltage maps (from CARTO-EP studies) using Vurtigo, an open source software. We then performed a quantitative analysis based on circumferential segments defined in the short-axis of MR images. Our results demonstrated a negative linear relation between bipolar voltage maps and T

1

maps within the first two mm of the endocardial surface. The results of our novel approach suggest that T

1

-maps combined with limited EP measurements can be used to evaluate the biophysical properties of healing myocardium post-infarction, and to distinguish between the infarct categories (i.e., dense scar vs. BZ) with remodelled electrical characteristics.

In recent years, the response of, especially, the right ventricle to intense exercise has gained increased interest. In this study, we use the CircAdapt model to evaluate the influence of inter-individual variation of pulmonary vascular resistance during exercise on cardiovascular hemodynamics. We modeled pulmonary vascular resistance as a nonlinear resistive module in which the resistance increases proportionally to stroke volume. We modeled inter-individual variation following normal random distributions. To evaluate the hemodynamic response to exercise, we computed pulmonary artery systolic pressure and end-systolic wall stress. With our modeling strategy, we were able to reproduce the phenomena observed clinically.

Despite the great potential of computational models, their clinical use is limited due to lack of clinical translation. It is, therefore, desirable to present and visualise simulation results as regular

imaging

data commonly used in clinical practice. The purpose of this paper is to present an implementation of a graphical user interface (GUI) for the lumped model CircAdapt. The GUI displays simulation results as:

1)

an animated schematic short-axis view of ventricular geometry,

2)

a Wiggers diagram relating pressure and volume curves with clear indication of the timing events within the cardiac cycle,

3)

a simulated Pulsed Wave Doppler echocardiographic image of mitral/aortic and tricuspid/pulmonary flow, and,

4)

a B-Mode echocardiographic image of short-axis ventricular geometry. We have modeled different physiological and pathological conditions to illustrate the applicability of the GUI: normal state, aortic stenosis and acute response to exercise with high pulmonary vascular resistance.

The dynamic deformable elastic template (DET) model has been previously introduced for the retrieval of personalized anatomical and functional models of the heart from dynamic cardiac image sequences. The dynamic DET model is a finite element deformable model, for which the minimum of the energy must satisfy a simplified equation of Dynamics. In this paper, we extend the model by integrating fiber constraints in order to improve the retrieval of cardiac deformations from cinetic magnetic resonance imaging (cineMRI). Evaluation conducted until now on cine MRI sequences shows an improvement of the recovery of the motion in images that present a low level of obvious rotation.

Sudden cardiac death is a major cause of death in industrialized world; in particular, patients with prior infarction can develop lethal arrhythmia. Our aim is to understand the transmural propagation of electrical wave and to accurately predict activation times under different stimulation conditions (sinus rhythm and paced) using MRI-based computer models of normal or structurally diseased hearts. Parameterization of such models is a prerequisite step prior integration into clinical platforms. In this work, we first evaluated the errors associated with the registration process between contact EP data and MRI-based models, using

in vivo

CARTO maps recorded in three swine hearts (two healthy and one infarcted) and the corresponding heart meshes obtained from high-resolution

ex vivo

diffusion weighted DW-MRI (voxel size < 1mm

3

). We used the open-source software Vurtigo to align, register and project the CARTO depolarization maps (from LV-endocardium and epicardium) onto the MR-derived meshes, with an acceptable registration error of < 5mm in all maps. We then compared simulation results obtained with the macroscopic monodomain formalism (i.e., the two-variable Aliev-Panfilov model), the simple Eikonal model, and the complex bidomain model (TNNP model) under different stimulation conditions. We found small errors between the measured and the predicted activation times, as well as between the depolarization times using these three models (e.g., with a mean error of 3.4 ms between the A-P and TNNP model), suggesting that simple mathematical formalisms might be a good choice for integration of fast, predictive models into clinical platforms.

Despite successful initial treatment during the acute phase of aortic dissections, long-term morbidity and mortality of type B aortic dissections is still a clinical challenge. Therefore, the importance of the assessment and understanding of potential variables involved in their long-term outcome, such as flow patterns and pressure profiles in false and true lumen and across tears.

Hence, we developed an equivalent electric 0D model mimicking a type B aortic dissection. The model was calibrated and validated using in-vitro experimental data from a pulsatile flow circuit. We assessed the variation of pressure profiles in the lumina and flow patterns across the tears as a function of changes in tear size and wall compliance.

We found a good concordance between the in-vitro experiments and the predictions from the lumped model.

Therefore, a 0D model of aortic dissection is feasible and offers potential to study pressures and flow pattern alterations in clinical conditions.