CE45 - Mathématique, informatique, automatique, traitement du signal pour répondre aux défis de la biologie et de la santé

Human and Animal NUmerical Models for the crANio-spinal system – HANUMAN

Human and Animal NUmerical Models for the crANio-spinal system

The main purpose of the HANUMAN project is to provide numerical models for human and non human primates. In particular, we will develop models <br />taking into account intracranial Pressure (ICP) versus arterial, venous and cerebrospinal fluid (CSF) flows oscillations in a realistic anatomical<br />craniospinal system developed from images acquired on human and marmoset monkeys (Callithrix jacchus), primate with a phylogenetical proximity to humans and usually used in preclinical studies.

Development ans comparison of Human and Animal NUmerical Models for the crANio-spinal system

The final application of such model will be, using flow measurements obtained non invasively by MRI, to evaluate the biomechanical properties of the human and monkey brain. With this new model we bridge the gap between animal models considered in preclinical studies, and numerical models developed for human, in clinical studies. <br />The first challenges are related to image analysis issues. In particular, it is mandatory to design image analysis methods adapted to marmoset 3T MRI. By contrast with human MRI, marmoset MRI visualizes structures of reduced size, with a low signal-to-noise ratio. <br />Second challenges concern the development of numerical models in 1D, 2D and 3D, their effective coupling for reproducing the blood flow in the vascular structures and the autoregulation process. Indeed, in the context of the craniospinal system, we need to study fluid flows and also interactions between fluids (blood, CSF) and structures. <br />Our models will aim at progressing in our understanding of many cerebral disorders in relation with blood and CSF circulations mainly present in neurodegenerative diseases. The expertise of clinicians in cerebral flows will be of importance, first in the acquisition of images, second in the analysis of images and numerical results, and finally for the transposition from the animal<br />model to the human one.

Imaging and image analysis for cerebrovascular networks and CSF :
MRI images acquired on humans will provide 3D information on vascular structures and ventricles, plus 3D + time information on CSF and blood dynamics. The quality and resolution of these images will lead us to develop methods and tools devoted to full 3D information extraction.
Our strategy for 7T images acquired on marmosets will consist in extracting structural information for designing no longer 3D numerical
models, but reduced dimension models, typically 1D models for vessels, and mixed 1/2/3D models for ventricles and SAS. This will allow us to develop reduced order (e.g. graph-based) numerical models that will lead to specific numerical simulation approaches.
Numerical simulation for biofluids : From in vivo 3D images, geometrical data and computational meshes will be constructed. They will be used for computing the coupled cerebral blood and CSF flows in the whole cranial-spinal structure with and without autoregulation. A multiscale modelling approach will be adopted. Local 2D/3D simulations will be embedded in a 1D/1D co-axial compliant network which couples blood and CSF flows both in the cranial and spinal compartments. The multiscale 3D/1D or 2D/1D models will take into account the fine local morphological aspects in some chosen parts and the network connective complexity, but also brain rheological modifications, to investigate the impacts on the dynamics. We will also investigate the challenging problem of pressure and parameters estimation based on MRI measurements.
Parameters estimation of the models : As some data cannot be measured non-invasively, characteristics that can not be assessed must be estimated by use of Kalman filtering for example.

We have developed and optimized the protocol for acquiring blood flow data in the marmoset monkey. These images are remarkable and represent a breakthrough. In particular, the analysis of these images seems to confirm that the aging mechanism of marmosets mimics that of humans, which was a question we wanted to address in this project. Indeed, the results obtained in marmosets showed differences between young and old subjects in the profiles
of the dynamics of cerebral blood flow, suggesting that the marmoset model converges to the human model. Data is being processed and analysed. Image processing algorithms are tested to segment arterial and venous networks from the TOF images and thus extract the information necessary for numerical simulations.
Regarding image acquisitions on humans, a phantom was performed under MRI for the purpose of quality control of flow sequences for pulsed flows. A novel ultra-rapid flow sequence has been validated for clinical use and then incorporated into the MRI protocol for blood flow and LCS. To visualize and process these new sequences, a dedicated software had to be developed. The originality of this software development is the possibility given by our treatment to access the influence of respiration on the dynamics of blood flow and the CSF in the craniospinal system.
The 1D numerical model coupling blood flow and cerebrospinal flow has
been successfully developed. Data extracted from the first marmoset images were used to simulate flows in a marmoset network.

Flow MRI measurements in intracranial CSF compartments on marmosets.
Computational 3D meshes for numerical simulations in 3D models.
Development of more efficient numerical methods for 3D models.
Comparison between young/old marmoset networks and.young/old human networks.
Parameters estimation of the 1D/3D models.

O. Merveille, B. Naegel, H. Talbot, N. Passat. nD variational
restoration of curvilinear structures with prior-based directional
regularization. IEEE Transactions on Image Processing,
28(8):3848-3859, 2019. 10.1109/TIP.2019.2901706

O. Balédent ISMRM 2019, Montréal.

O. Balédent ICP 2019 Louvain.

Bazzi F, Rodriguez-Callejas JDD, Fonta C, Ahmad Diab A, Amoud
H, Falou O, Mescam M, Basarab A, Kouam´e D. 2019. Brain
Segmentation from Super-Resolved Magnetic Resonance
Image. Fifth International Conference on Advances in
Biomedical Engineering (ICABME), October 17-19, Tripoli
Campus, Lebanon : 4 pages.
doi.org/10.1109/ICABME47164.2019.8940281

N. Passat. Component-trees: Structural and spectral extensions.
Workshop on Digital Topology and Mathematical Morphology
(DTMM) 2019.

P. Mollo Freefem++ Days Décembre 2019 “Reduced basis with
Freefem++. Application to Stokes equations”
Référence du formulaire : ANR-FORM-090601-02-02 7/9

P. Mollo VPH2020 Août 2020 “A reduced basis method for computing
cerebrospinal fluid hydrodynamics”
Poster Student Award

The relevance of numerical simulations in the field of medicine is now obvious. They give access to information that could not be obtained \emph{in vivo} or non-invasively in humans. Animal models also constitute a means of obtaining such information. Nevertheless, their use remains limited by ethical issues, but also by uncertainties related to the compliance between humans and animals.

In this context, there exist very few numerical approaches devoted to the animal.
However, such approaches would allow for tackling ethical questions by minimizing experiments on animals.
Indeed, coupling \emph{in vivo} technics on animals ---respectful of ethics--- with \emph{in silico} approaches would open the way to handling complex physiological parameters and studying compatibilities between human and animal models.
Then, we propose to design numerical models dedicated both to animal and human.
In particular, we will develop numerical models of the craniospinal system (liquid flows: blood and cerebrospinal fluid -- CSF) on human and marmoset monkey (\emph{Callithrix jacchus}), a small primate frequently used in preclinical studies due to its phylogenetic proximity to the human.

Indeed, the brain has not been extensively studied from the side of its fluid dynamics. However, recent studies show that blood flows and CSF hydrodynamics (in particular, intracranial pressure oscillations) have a crucial role in the correct functions and perfusion of the brain. These considerations motivate research work on interactions between blood and CSF.

By contrast to static tissue studies, that can be carried out \emph{ex vivo}, the blood flow and CSF mechanics have to be studied \emph{in situ}, on living subjects. For ethical and technical reasons, magnetic resonance imaging (MRI) constitutes a relevant tool since it allows for the non-invasive acquisition of images. In addition, recent progress in MRI now enables to observe the flows in 3 dimensions + time, for quantifying \emph{in vivo} oscillating CSF and flowing blood.

Images will be acquired on humans and marmosets. The extraction of various kinds of information (geometry of vessels and CSF compartments; networks structure; velocimetric data) from both 3 and 7 Tesla MRI images ---never used for such purpose--- will require to develop specific methods and tools dedicated to marmoset images, that induce specific challenges (resolution, signal-to-noise ratio).

For similar reasons, numerical models especially designed for blood and CSF simulation will be proposed, in order to study the fluid--fluid--structure coupling and to look for physical parameters of interest. Models of different dimensions will be considered: 1D models for global behaviour analysis in complete networks; 2D models for investigating intracranial pressure autoregulation process; 3D models for observing accurate behaviour of flows.

The complexity of the involved physiological mechanisms makes the determination of important physical parameters a hard task. Numerical models then constitute an exploratory tool of high interest. Numerical animal models will be finally correlated to human models, with the purpose to assess the validity of the proposed hypotheses and the potential transferability from animal to human.

This project, highly pluridisciplinary, gathers mathematicians, computer scientists, biologists and medical doctors.
The models will be developed and released as open-source softwares, thus guaranteeing their wide diffusion within the scientific communities.
They will contribute to improve our understanding of mechanisms related to cerebral pathologies.

Project coordination

Stéphanie Salmon (LABORATOIRE DE MATHÉMATIQUES DE REIMS - Université de Reims Champagne Ardenne -)

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.

Partner

IMFT INSTITUT DE MECANIQUE DES FLUIDES DE TOULOUSE
CerCo CENTRE DE RECHERCHE CERVEAU ET COGNITION
CHIMERE
LMR - URCA LABORATOIRE DE MATHÉMATIQUES DE REIMS - Université de Reims Champagne Ardenne -

Help of the ANR 439,958 euros
Beginning and duration of the scientific project: November 2018 - 48 Months

Useful links

Explorez notre base de projets financés

 

 

ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter