NOTE: this position listing has expired and may no longer be relevant!
Context:
The proposed PhD thesis belongs to a project (MRI_quantif) gathering the hydrodynamic group of LHEEA Lab. of Ecole Centrale Nantes, the signal and image processing group of IRCCYN Lab. of Ecole Centrale Nantes, and the radiology group of the Thorax Institute of the University Hospital of Nantes, under funding of the Pays de la Loire French region.
The final objective of this project is to quantitatively improve extraction of biomarkers from MRI (Magnetic Resonance Imaging) measurements of neuro-cardio-vascular flows. Biomarkers are biological, physical or chemical quantities whose measure is of clinical interest.
MRI measurement of flows is getting more and more effective and accurate. Recently, a new generation of MRI devices for flow visualization and measurement arose, known as 4d-flow (or 7D) MRI, which enable 3D representation of flow evolutions in time. However, images/movies produced are the result of several combined techniques (magnetic resonance sequence choice, signal and image processing, post-processing, etc.) making difficult an intrinsic assessment of their quality.
To provide quantitative assessment and improvement of the measured/visualized quantities, the project is decomposed into two main tasks: 1/ super-resolution: improving the quality and resolution of images through the use of image processing and fluid mechanics knowledge; 2/ validation: quantitatively evaluating the quality and flaws of 4D flow MRI measurements and super-resolved images from task 1 through comparison to other reference techniques used in hydrodynamics, especially to dedicated test bed experimental results with controlled parameters.
Eventually, the quantification and improvement of MRI flow measurement will enable a better definition of existing biomarkers of clinical interest, and to propose others.
Objectives:
The objectives of the proposed PhD thesis are:
i/ Provide means to validate data from standard preclinical and clinical MRI measurements and post-processed super-resolved data.
ii/ Quantify existing biomarkers from the super-resolved data obtained with the algorithms developed in task 1 of the project, and propose others if relevant.
Regarding the first objective, the PhD student will define, setup, run and analyse an experimental test bed enabling validation. A MRI-compatible phantom will be created. Flows in this phantom will be measured both with 4D-flow MRI and an accurate reference technique, e.g. LDV (Laser Doppler Velocimetry). Cross validation of these two techniques will be achieved on increasingly complex test cases: starting from stationary flow in simple geometries to more and more complex pulsatile flows mimicking human blood flows such as the one in the aorta. LDV measurement calibration will first be performed separately, to ensure the quality of this reference technique. Then, a complete MRI-compatible pulsatile flow setup will be developed, and MRI measurement quality will be quantified against reference LDV ones. Quality of super-resolved MRI data will then be assessed, and this assessment will serve as validation and improvement basis of the algorithms developed in task 1 of the project. Finally, when the whole measurement/super-resolution/quality assessment process will be validated, MRI measurements will be performed, super-resolved and analysed on real human blood flows.
Depending on the results, two other aspects will be further considered. First, in real human flows, the vessel boundaries are compliant and hydroelastic effects occur. These effects will be neglected in a first approximation but it will be quantified whether this assumption holds. If not, compliance will be introduced in the whole measurement/super-resolution process, which will clearly be an additional difficulty to tackle. Second, it is likely that reference flow measurements alone will not be sufficient to serve as validation of the super-resolution algorithms in every configuration, especially in geometrically complex ones as are in the human body. In that case, numerical simulation will be used, alone or in combination with measurements (e.g. standard MRI measurements in inlet/outlet sections), to provide additional validation resource. Again, accounting for the compliance would increase a lot the difficulty of such simulations in that case.
In parallel, the PhD thesis will be dedicated to the evaluation of existing biomarkers, and proposition of new ones. Hence, during task 1 developments and task 2 validation process, focus will be made on the measured quantities which are and/or could be used as biomarkers. Among them, the viscous stresses at the wall (wall shear stress) will be of first interest. This quantity is a challenging target since it involves derivatives of the blood velocity field where it varies most, i.e. along the vessel boundary which itself needs to be identified through segmentation between vessel wall and vessel core on relatively low spatial resolution images. Good image resolution there is thus both difficult and crucial. Quantitative evaluation of these target biomarkers will thus lead the choices made all along the project. Depending on the findings made, the results could also lead to proposition of new relevant biomarkers.
Supervision:
The PhD thesis will be supervised by Prof. David Le Touzé and Félicien Bonnefoy from LHEEA fluid dynamics lab. of Ecole Centrale Nantes and Prof. Jean-Michel Serfaty from the Thorax Institute of Nantes University Hospital.
David Le Touzé, david.letouze@ec-nantes.fr
Félicien Bonnefoy, felicien.bonnefoy@ec-nantes.fr
Jean-Michel Serfaty, jeanmichel.serfaty@chu-nantes.fr
Profile:
Applicants shall hold a Master/M.Sc. degree, preferably in mechanical engineering (ideally in fluid mechanics or bio-mechanics) or applied mathematics.