Observing cerebral blood flows in the brain with MRI is a public health challenge. It requires specific image modalities, and depends on new technology at the frontier between medicine, physics, mathematics and computer science. The objective of this project is to propose new tools to better comprehend these complex flows.
Imaging blood flows (a.k.a. angiography) is used for the diagnosis, follow-up and treatment of various conditions affecting the vascular network, in particular in the brain. Improved image analysis precision implies better outcomes for various procedures and interventions. To this end, we identify three distinct but linked challenges: (1) the development of new, intelligent angiography analysis tools; (2) a better understanding of blood flows in the brain at the fluids dynamics level; (3) mastering physical image acquisition procedures for angiography. This project aims at proposing answers to these three challenges simultaneously. We have assembled a multi-disciplinary team, whose goal is to provide a complete angiographic processing pipeline, from the image acquisition to the simulation of realistic virtual angiography images. We are developing high-resolution, computerized analysis tools for extracting the geometry of blood vessels, for numerically simulating blood flow in these networks, and physical modelling tools that simulate the visualization techniques of these <br />flows by magnetic resonance imaging. Beyond expected theoretical and practical improvement in the individual techniques employed in this research (computer science, mathematics, physics), the synergy between these disciplines is expected to deliver improved understanding in the internal mechanisms of the cardio-vascular and cerebral networks. This should allow us, as a perspective, to deliver significant progress in medicine and to tackle major public health challenges.
The three main questions we seek to propose answers to are the following: (1) how do we extract the three-dimensional structure of the cerebral blood vessels from MRI data? (2) How do we simulate blood flow in these vessels? (3) How do we simulate the way MRI senses these flows to translate them into visual data? Answering the first question requires developing new image processing techniques that reproduce human perception, or even improve upon them in the case of complex, 3D image data as in MRI angiography. So-called segmentation procedures, i.e. the classification of visual features in regions of interests are notably employed. Answering the second question requires utilizing the geometric structures extracted in the previous question, and employ methodologies developed in fluid mechanics, particularly involving complex space-time dynamical equation systems, requiring the use of advanced numerical tools and massive utilization of supercomputing facilities. Answering the third question necessitates to put together and compare the numerical simulations performed in the second question with the added information needed for MRI simulation (e.g. the magnetic response of the blood) with real MRI machine output. This will allow us to experimentally determine the importance of various physiological, physical and environmental parameters that influence the reconstruction of angiographic images. In turn, we can use these results to improve the methods for answering the first question.
At the end of the project, many advances are expected, both at the fundamental and technological level. Research-level developments that are useful to the community of scientists in the domains involved in this study will be translated into freely available software tools. This will be put into place via an industrial and academic collaboration thanks to the implication of the company Kitware in our consortium. Kitware is a world leader in the medical image processing and image visualization software, as the developers of the free ITK and VTK toolkits. Other outcomes of the project such as three-dimensional vascular networks, blood flow numerical simulations, virtual MRI images, etc., will allow other scientific communities to continue work initiated in this project, and so to promote multi-disciplinary approaches for solving the underlying medical challenges not initially considered in this project, e.g. the development of better tools for the better diagnosis of various brain vascular conditions, such as stenosis, aneurisms, arterio-venous malformations, etc. These conditions are increasingly encountered in industrialized countries.
Understanding blood flows in the brain constitutes an unsolved challenge. Overcoming this will almost certainly uncover even more complex problems, such as the global analysis of the human cardio-vascular system and its interaction with other bio-fluids, such as the cerebrospinal fluid. Another challenge is the modelling of the micro-vascular structure of living tissues, which are currently modelled at a very rough level. These long-term perspectives will necessitate new conceptions in multi- and inter-disciplinary scientific research and innovation, which could result in the development of novel therapeutic strategies.
As in every research project, scientific production will be translated into the creation of theoretical knowledge and practical methodologies. The writing of articles, as well as the production of software and datasets will provide these to the entire sci
In the last 20 years, progress in medical imaging has led to the development of modalities devoted to visualise vascular structures. These angiographic images progressively proved their usefulness in various clinical applications, in particular for cerebrovascular issues. However, these cerebral angiographic data are complex to analyse. This has motivated, since the mid 90’s, the proposal of several image processing tools for vessel analysis. Unfortunately, contrary to the morphological brain image analysis, for which synthetic (i.e., virtual) images and associated ground-truths (segmented data) are available (e.g., BrainWeb), there are not such data in the case of cerebrovascular images.
The interdisciplinary program starts from real MR angiographic data to finally lead to the generation of virtual MR angiographic data. During this process which leads to these simulated data, realistic 3D (anatomical) and 3D+t (hemodynamic) models –providing ground-truth for the virtual MRA images– are obtained.
The purpose of this project is to develop a pipeline for the generation of virtual Magnetic Resonance Angiographies (MRA) of the human brain, associated to their anatomical (3D) and hemodynamic (3D+t) models (providing a ground-truth).
The simulated data and ground-truths are currently not available for cerebral vascular networks. This results in a lack of common development, validation and comparison framework in the research fields related to vessel analysis
The project needs to consider the following issues:
• The handling of inter-individual cerebrovascular variability, in the context of anatomical model generation.
• The numeric simulation of blood flows in complex geometries and in multiphysic and multiscale frameworks.
• The accurate simulation of the physical processes involved in MRA acquisition sequences in order to finally obtain realistic virtual angiographic images.
In order to do so, the successive steps are considered:
• Extraction of vascular volumes from real MRA images.
• Generation of 3D vascular models from these data.
• 3D+t simulation of blood flow in complex (arterial and venous) models.
• Simulation of MR acquisition of angiographic data from these 3D+t models.
The interaction between and within these steps requires strong interdisciplinary collaborations between computer science, applied mathematics, physics and medicine. The project combines all these expertise within the academia partners which are completed with the world industrial leader in open-science (open-source and open-data), Kitware, in bio-medical software and associated data infrastructures. In particular we develop upon the standard open frameworks developed by Kitware to efficiently deal with the issue of large data storage, handling, and computation to finally and effectively lead to operational data and software resources.
Two major deliverables are set for this project: (i) the obtained data (virtual images and associated 3D/3D+t models) will be finally available on the web, via a server accessible to the whole medical community and (ii) the methodological solutions developed at each step of the project will lead to software tools relying on standard (open source) software resources, then being also fully available to the medical image analysis community.
The common and reliable framework for the design, calibration, validation and comparison of angiographic image processing and analysis (filtering, segmentation, quantification, etc.) developed in this project will be an invaluable accelerator for the academic and industrial partners involved, their research and innovation works, thus leading to technological and medical improvements. The socio-economic importance of the research proposed is highlighted by the fact that vascular pathologies are one of the main causes of morbidity and mortality in the Western world, and thus constitute a crucial public health issue.
Monsieur Nicolas Passat (Université de Reims Champagne-Ardenne - Laboratoire de Mathématiques de Reims) – firstname.lastname@example.org
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Kitware Kitware SAS
UNISTRA - LSIIT Université de Strasbourg - Laboratoire des Sciences de l'Image de l'Informatique et de la Télédétection
UNISTRA - LINC Université de Strasbourg - Laboratoire d'Imagerie et de Neurosciences Cognitives
URCA - LMR Université de Reims Champagne-Ardenne - Laboratoire de Mathématiques de Reims
ESIEE - LIGM ESIEE-Paris - Laboratoire d'Informatique Gaspard-Monge
UJF - LJK Université Grenoble 1 - Laboratoire Jean Kuntzmann
Help of the ANR 966,009 euros
Beginning and duration of the scientific project: December 2012 - 48 Months