Optimization of the reconstruction of viscoelastic parameters obtained by a Non-Linear Inversion method on in vitro Magnetic Resonance Elastography data

Subject: Optimization of the reconstruction of viscoelastic parameters obtained by a Non-Linear Inversion method on in vitro Magnetic Resonance Elastography data

Duration: 6 months (from February)

Location: MAGICS team of CREATIS laboratory. Bâtiment Léonard de Vinci Campus La Doua, 21 avenue Jean Capelle, 69621 Villeurbanne, France

 

Description and aim of the internship:

This project falls within the field of medical imaging, with its application area being Magnetic Resonance Elastography (MRE). This imaging technique allows for mapping the biomechanical properties of soft tissues such as the brain or liver using images acquired through Magnetic Resonance Imaging (MRI)1. There is currently no imaging method capable of directly mapping biomechanical properties, such as elasticity and viscosity. Thus, MRE involves three steps. First, using an external actuator such as a pneumatic or piezoelectric system, the tissue of interest is mechanically stimulated, inducing the propagation of a shear wave within the tissue. The generated displacements are then measured using a specific MRI sequence, yielding a series of volumetric images depicting wave motion in three orthogonal spatial directions. Finally, these wave images are processed using a dedicated numerical reconstruction algorithm which enables the identification of the biomechanical properties of the tissue under investigation, based on its response to a controlled stimulation. The commonly used algorithm is based on the inversion of the wave equation2. More complex algorithms, such as the Nonlinear Inversion (NLI)3 method, are currently being explored. NLI is based on meshing the tissue and defining the mechanical properties at each node of the mesh. Solving the inverse problem involves finding the mechanical parameters of the mesh that allow simulating the propagation of a wave minimizing the differences with the acquired wave images. NLI could thus offer a better characterization of biomechanical parameters from multi-frequency MRE4 data, considering factors like tissue behavior laws and tissue anisotropy. An overview of MRE and NLI reconstruction method is provided in Figure 1.

 

The proposed research project specifically aims to study the performance of the NLI reconstruction algorithm compared to the algorithm traditionally used. Images will be acquired on a preclinical MRI5 where various parameters such as field of view, image resolution, and wave frequencies are significantly different from those encountered in clinical settings, where the NLI method has been developed and tested. Multi-frequency MRE data will be acquired using a preclinical MRI on the PILoT imaging platform, using samples created to simulate various in vivo scenarios. This internship will be jointly supervised by E. Van-Houten from the University of Sherbrooke (video conference meetings are planned) and may benefit from the use of the 'Calcul Canada' infrastructure for performing numerical computations.

More specifically, the objectives of the internship will be as follows:

  • Create phantoms with inclusions of different sizes and stiffness, using different materials.

  • Perform multi-frequency MRE experiments on a preclinical 7T MRI scanner

  • Optimize the parameter settings for the reconstruction of biomechanical properties using the NLI method: subzone size, mesh resolution, boundary conditions, etc.

  • Reconstruct the viscoelastic properties of these samples using the NLI method and the reference method, and conclude on the potential of the NLI method.

  • Verify and compare the viscoelastic behavior of agarose and plastisol according to rheological laws.

Candidate profile:

Skills in biomechanics and MATLAB programming for data processing are required. Knowledge in MRI physics would be appreciated and an affinity for experimental measurements is desirable.

 

References:

1. Asbach P, Klatt D, Hamhaber U, et al. Assessment of liver viscoelasticity using multifrequency MR elastography. Magnetic Resonance in Medicine. 2008;60(2):373-379.

2. Oliphant TE, Manduca A, Ehman RL, Greenleaf JF. Complex-valued stiffness reconstruction for magnetic resonance elastography by algebraic inversion of the differential equation. Magnetic Resonance in Medicine. 2001;45(2):299-310.

3. Three-dimensional subzone-based reconstruction algorithm for MR elastography, EEW Van Houten et al., Magnetic Resonance in Medicine, 45(5), pp. 827—837, 2001.

4. Viscoelastic power law parameters of in vivo human brain estimated by MR elastography, J Testu et al., Journal of the mechanical behavior of biomedical materials, 74, pp. 333—341, 2017.

5. P. Sango-Solanas et al. Parametrization of multi-frequency non-linear inversion reconstruction on in vivo mouse liver magnetic resonance elastography data, ITEC 2024.

 

Supervision: 

Pilar SANGO-SOLANAS and Kevin TSE VE KOON: MCUs at Université de Lyon 1, Laboratory CREATIS

Elijah VAN HOUTEN: Professor at Université de Sherbrooke

 

Application: 
Send a CV and a cover letter to  Pilar.Sango@creatis.univ-lyon1.fr and kevin.tsevekoon@creatis.univ-lyon1.fr