Achievements

Overview of the project activities during the second reporting period and main outcomes achieved with the second year of activities

The project has entered in its core phase. In fact, besides the completion of the data collection process, propaedeutic to the development and validation of the modelling and simulation tools, important goals have been achieved on the modelling side, leaving now a year to clinically validate the results, while refining the interfaces to guarantee the maximisation of the usability of the implemented tools within the clinical environment.

Clinical Activities

During the Second Reporting period, the clinical activities focused mainly on two specific tasks: the completion of the data collection process and the clinical validation of the modelling results. With regard to data collection, the clinical partners have managed to make some good progress with the enrolment (which is almost completed) of both patients with aortic valve disease and with coarctation of the aorta. To complete this task, however, during the Second Biannual Meeting (held in Graz in March 2015) it has been deemed necessary to postpone the end of the data collection process for a couple of months, to achieve the foreseen targets, particularly with regard to CoA patients. In fact, the Consortium experienced some problems in recruiting the expected numbers of CoA patients, mainly due to the low rates of recoarctation after initial surgical coarctation repair observed in the UCL patient population, making it difficult for UCL to be able to collect the foreseen 20 cases of coarctation requiring a stenting intervention. This circumstance and the subsequent mitigation strategy adopted to overcome the issue are further addressed in the report of WP2. Besides the data collection process, a Matlab software for the analysis of pressure curves was further developed, whilst aorta segmentations, ventricle segmentations, and myocardial wall segmentations were completed. Furthermore, flow rates analysis was performed on all enrolled patients. As concerns clinical validation, the work is proceeding as planned. During the second reporting period, the following objectives were achieved:

  • the virtual valve treatment procedure for the clinical validation of the model-based predictions has been elaborated, and some AVD patients were compared with regard to predicted post-treatment geometry and hemodynamics
  •  Diagnosoft HARP software, which calculates strain from tagging MRI sequences, is under testing.
  • Energetic aspects of the heart work were calculated for 15 CoA patients enrolled in DHZB.
  • Stenting tool developed by MEVIS was verified against post-treatment segmentation done in DHZB
  • Area compliance assessment as a parameter for clinical validation of the fluid-structure interaction model was selected and the assessment procedure was elaborated.

Data Infrastructure and Information System development

The main achievement regarding the development of the CARDIOPROOF technical infrastructure has been the activation of the independent gateway at DHZB, connecting with the Infostructure. Nowadays, DHZB is the first node connected to MD-Paedigree that is not located inside OPBG, and this surely represents a great advancement in terms of proof of concept for the technology. To succeed in such fundamental task, a close cooperation between P9 Gnúbila and the IT technical responsible at DHZB was necessary, to integrate the CARDIOPROOF Infostructure within the security framework of DHZB. Subsequently, the data already collected and uploaded on the TrialConnect system (already activated during the first year of the project) should be integrated within the CARDIOPROOF/MD-Paedigree Infostructure. Finally, with regard to the usability of the tools, testing activities on the Fraunhofer MEVIS virtual stenting software have been successfully completed by the clinicians at DHZB, proving its usability in a clinical environment.

An outlook on the modelling activities

As already recalled in the first Annual Report, CARDIOPROOF encompasses different modelling activities, distributed in different workpackages. The outcome of these WPs will be eventually integrated in a comprehensive modelling tool with different features, capable to provide the clinicians with a complete and in-depth simulation of the treatment opportunities and outcomes. The paragraphs below anticipate some of the results performed in the second year of activity in each modelling domain. These results are presented in detail in the subsequent WPs section.

The mechanics- Computational Fluid Dynamics (CFD) interface and finite element mesh generation pipeline

MUG, building on the results achieved during the first reporting period, focused its activities within WP4 in the parametrisation of the electrophysiological model, the mechanical model, and the coupling of these two models with each other. Furthermore, the coupling of the mechanical model with lumped models representing the cardiovascular system has been performed.Senza titolo1

In particular, parametrisation strategies were implemented (T4.3), in order to generate patient-specific electrical activation and repolarisation sequences that will serve as input for the finite elasticity model used to compute mechanical deformation. In order to compute in an efficient way the electrical activation sequence, a solver for a low-dimensional phenomenological Eikonal model was developed. Additionally, to approximate the mechanical deformation of the heart in a patient-specific manner, parametrisation and data assimilation strategies were implemented (T4.4).

Finally, to overcome the huge computational load and subsequent execution time of the simulation of cardiac electro-mechanics with biophysically detailed models, multigrid preconditioners for cardiac mechanics will be integrated in the implemented solver framework, while the entire FE pipeline will be ported for GPU execution (T4.5). This should enable significant reductions in execution time, making the time scales needed for the simulation of a full cardiac cycle compatible with the envisioned clinical workflow.

Software tools for the integration of the computational models of the morphology and the 4D velocityencoded cine magnetic resonance imaging (VEC MRI) information

Building on top of the approaches implemented during the first reporting period, a new SPH filling algorithm has been developed by ESI, in order to improve the quality of the initial particle distribution. Additionally, another improvement has been made, in the form of an additional simulation in which the initial particle distribution expands until equilibrium with respect to the contact with

Senza titolo2the walls of the LV and aorta. These improvements, together with minor optimisation of the aortic valve opening and of the start of the outflow into the aorta, led to the obtainment of realistic results in the simulation of the blood flow. Furthermore, a software to generate a numerical valve model for a given FE model of the heart has been implemented. Thanks to the valve opening model, the simulation appears more realistic, compared with the simulation with a sudden opening. Finally, a Windkessel option was implemented as part of the SPH outflow boundary condition. The results for the simulation with the Windkessel model are consistent with the results of a simulation with a corresponding outlet pressure, indicating that such a model is working correctly. Most of the implementation developed in this reporting period will also contribute to the future clinical use of the tools.

Virtual Stenting Software

During the second reporting period, P5 Fraunhofer MEVIS has further refined its Virtual Stenting software, which was already rather definite at the end of the first year. During the second year of activity, in order to provide the clinical end-users with usable virtual stenting and haemodynamic outcome simulation, major progress has been achieved in the generation of the input for the CFD simulation, which has been implemented in a software prototype. In particular, the image-based CFD simulation has been integrated with the image analysis workflow and validated through comparison of the resulting pressure gradients with clinical catheter measurements and other CFD simulations. Another major achievement is the validation of the virtual stenting software,

Senza titolo3which has been performed through the reproduction of post-interventional geometries by interactively changing the pre- 11 Vorticity magnitude visualization at peak systole. Aortic valve in grey. interventional geometries of 8 cases with treated coarctation of the aorta. Furthermore, initial simulations have been performed on the clinical cases made available on the TrialConnect platform, with good results. The entire validation process is performed in close cooperation with WP8. Furthermore, in line with the integration concept proposed in WP3, a prototypical server-based infrastructure has been developed, which provides a web interface for the interactive workflow steps. Also, this web tool has been preliminarily tested by the clinicians at DHZB.

Computational tool for computing the pressure – difference field from 4D Flow data

SAG has continued, during the second year of the project, with the implementation of tools for deriving clinically relevant parameters for the assessment of aortic dilatation from non-invasive data.In particular, the work in WP7 focused on the realistic coupling of fluid and solid models for FSI analysis, on the basis of the work already performed during the first year. The implementation of tools for the integration of the aortic vessel walls as well as thin, moving structures (such as the aortic valve leaflets) into the already developed FSI framework, represents a key achievement. This should enable more accurate simulations in realistic scenarios. Senza titolo4On another note, other progress has been made on the relative pressure estimation in the aorta from MRIbased 4D flow data, introducing, in particular, an approach based on a reduced-order blood flow model. During the second half of the year, a specific effort has been put in place for the design and implementation of a fully-automatic estimation method for local/regional mechanical vessel wall properties from non-invasive data. This will enable personalised reduced-order FSI simulations. Finally, an early prototype for clinical application (which encompasses patient browsing, visualisation and data interaction capabilities for patient images and other measurements, as well as computed/estimated results/parameters) has been developed, in the form of an intuitive UI.

 

Overview of the project activities during the first reporting period and main outcomes achieved with regard to the activities performed in the first year of the project

It can be said that the project is overall on track and that the achieved results have been satisfactory and will be a good basis for further development toward the main goals of the project.

Clinical Activities

The initial activities on the clinical side comprise a preliminary approval of the study granted by the local Ethical Committees in Rome, London and Berlin, with the parallel preparation of specific informed consent for patients or parents. After these necessary propaedeutic steps, at the very beginning of the project, a data acquisition protocol was discussed (also during the face to face meeting in Berlin, the Kick-Off Meeting), developed and agreed amongst the clinical partners, including relevant inclusion and exclusion criteria for the patients involved in the project. Thus, the data collection process started and, at the end of this reporting period, a total of 54 patients have been enrolled within the three clinical centres. The data collected allowed the first simulations by modelling groups, with particular regard to the validation of the first tools (stenting tool as developed by MEVIS and pressure mapping tool as developed by SAG). Furthermore, the data gathered on the flow rates in the aorta have been made available for the simulations planned in WP5, 6 and 7. Finally, the statistical analysis comparing pre- and post-treatment flow rates has been accomplished.

 Usability Engineering, Data Infrastructure and Information System development

The development of a technical infrastructure to host, share and analyse the patients’ dataset is one of the fundamental and propaedeutic objectives of CARDIOPROOF. TO achieve this major goal, P9 Gnúbila has closely cooperated with the clinical and technical partners to put in place the first prototype of the above mentioned IT infrastructure, building upon the Infostructure implemented within the ongoing FP7 project MD-Paedigree, of which the CARDIPROOF one is intended to be an extension (aiming to add at least one node to the system). This also allowed to optimise time, effort and funding, making it possible to concentrate on the specific aspects and needs of the project. In fact, since most of the solutions provided so far were already implemented for MD-Paedigree, it has been possible to significantly reduce costs for hardware, software and maintenance. In Cardioproof, the data infrastructure and information system is based on Gnúbila’s FedEHR product and it is composed of two main components:  A backend, composed by all the Web Services being the ground of the platform with the addition of external applications; this has been inherited from MD-Paedigree and is the foundation of the system. 9  A frontend, offering a user-friendly web interface allowing user communities to easily interact with the platform. It is the main entry point to the system, and has been built by extending the MDPaedigree front end. It exposes all the Cardioproof applications and tools through a common and simple web-based interface. Some of the activities performed during the first year have been also devoted to integrate the above described Infostructure and the IT systems already operating at P2 DHZB. One of the latest development has been the discussion upon the possibility to make us of the ENSEMBLE system, which can extract, process and route data to target, which will be used to access DHZB data and to bridge its database with the FedEHR system. On a different level of the work, P5 MEVIS concentrated on usability and user requirements issues, analysing for this purpose (with the cooperation of the clinical partners) the current status of therapy planning, decision workflow, and usage patterns in clinical practice. To achieve an in-depth understanding of the clinical need and context, the entire patient pathway and the clinical workflows and requirements were investigated, with specific visits to the clinical centres themselves, which were conducted in the first year. The following steps will also include a usability test of the stenting simulation software. The steady involvement of the clinicians in the creation of workflows and design of the software interface will eventually guarantee a better experience with the software and the higher level of usability.

 An outlook on the modelling activities

CARDIOPROOF encompasses different modelling activities, distributed in different Work Packages, which will be eventually integrated in a comprehensive modelling tool with different features, capable to provide the clinicians with a complete and in-depth simulation of the treatment opportunities and outcomes. The following paragraphs will provide a brief explanation of the work performed in the first year in each modelling domain.

The mechanics- Computational Fluid Dynamics (CFD) interface and finite element mesh generation pipeline

MUG activities in the first year have been devoted to the definition of a technical standard for data exchanging between the electro-mechanical simulation output, as produced by the simulation codes used at the MUG, and the CFD codes used in other work packages. To perform the simulation, 3D volumetric Finite Element (FE) meshes of the heart were used. This allowed the inner surfaces of the heart to be extracted from the 3D volumetric mesh. The overall process is illustrated in Figure 1. Surfaces relevant for the definition of CFD boundaries are labelled in the volumetric mesh (Step 1), extracted for all time steps computed in the electro-mechanical simulation run (Step 2) and used as input in the CFD simulation (Step 3).

Senza titolo1

Furthermore, MUG worked on the development of a processing pipeline fortranslating segmented clinical image data on 4D cardiac anatomy into finite element models of the ventricles and the attached vessels. These models can be used in computer simulations of electrical activity and concomitant mechanical deformation of the heart. This work contributes to the overall objective of Work Package 4, which is to develop, apply and validate modelling methodology for performing patientspecific in silico simulations of ventricular electro-mechanics.

Software tools for the integration of the computational models of the morphology and the 4D velocityencoded cine magnetic resonance imaging (VEC MRI) information

Senza titolo2

ESI have defined a technical standard, which allows exchanging data between the morphology of the ventricle and aortic valve, using the segmentation of the images acquired by the clinical partners and related MRI information. This work contributes to the haemodynamical modelling of the ventricle and aortic valve. For the simulation of the blood flow in the cavities, the SPH solver within the Virtual Performance Solution (VPS) software is going to be used.

Virtual Stenting Software

MEVIS has achieved, during the first year of work, one of the more interesting and results, having developed and refined the virtual stenting software.

The importance of such an achievements become clearer when associated with the relevant clinical need: the treatment of the aortic coarctation, a narrowing of the aorta in the region of the transition between the aortic arch and the descending aorta, that occurs in about 7% of all congenital heart defects.

Senza titolo3

For aortic coarctation, the balloon angioplasty with stenting is the minimally-invasive recommended therapy (specifically for patients with a systolic pressure gradient of more than 20 mmHg). However, the optimal placement of the stent is not a simple task. Here come in support the Virtual Stenting Software, which make it possible to simulate treatment options prior to the intervention, thus helping the clinicians to choose the optimal stent placement, which guarantees a better outcomes in term of changes in aortic diameter  and better post-interventional hemodynamics.

Computational tool for computing the pressure- difference field from 4D Flow data

SAG developed, during the first period of activity, a two-step method useful to compute the relative pressure within the aorta noninvasively from magnetic resonance imaging (MRI). In the first step, the blood velocity acquired by phase contrast MRI is reconstructed from the measurements, as it is not discretely incompressible and suffers from various artefacts, especially near the vessel walls. The second step takes the reconstructed velocities to compute the time-varying 3D relative pressure within the aorta. This technique has been applied to three patients from the CARDIOPROOF project, delivering sound results even with imperfect data. Moreover, the whole pipeline will be validated using the coarctation cases, also acquired within the project, which include invasive pressure measurements. That way, it will be possible to directly compare the simulated pressures with the measured ones. Relationships between treatment outcomes and simulation results from this database will be investigated.

Senza titolo 4