PhD position - Pressure Loss in Arterial Stenoses, Measurement by Doppler Ultrasound and Artificial Intelligence
Recrutement: 
Recrutement en cours/passé: 
Recrutement en cours
Periode: 
2023-2026
Contact: 
Carole Frindel, MYRIAD team at CREATIS: machine/deep learning, informatics, medical imaging. carole.frindel@creatis.insa-lyon.fr homepage: https://sites.google.com/view/carole-frindel Damien Garcia, ULTIM team at CREATIS: fluid dynamics, numerical analysis, ultrasound imaging. damien.garcia@creatis.insa-lyon.fr homepage: www.biomecardio.com

Pressure Loss in Arterial Stenoses, Measurement by Doppler Ultrasound and Artificial Intelligence

Keywords: computational fluid dynamics, machine/deep learning, ultrasound imaging, in vitro experiments

Expected skills: fluid dynamics, programming in Python, knowledge of a machine/deep learning library, experimentation 

Project duration: 3 years funded by “école doctorale MEGA”

Clinical context: This Ph.D. project involves the evaluation of the severity of vascular narrowing (stenosis). The measurement of pressure loss associated with vascular stenosis is usually performed by catheterization, an invasive medical procedure consisting of inserting a thin tube into the bloodstream.

Objective: The objective will be to develop and experimentally validate a new non-invasive technique for the measurement of pressure loss in vascular stenoses. This technique will be based on ultrasound imaging and artificial intelligence and will meet the clinical requirements.

Methodology: The pressure loss in vascular stenosis can be written as a function of the flow rate [1]. The parameters of this functions are scalars that mostly depend upon vascular geometry. The goal will be to estimate these three parameters from the vascular geometry determined by ultrasound imaging. The estimator will be an artificial intelligence (AI) based model.

The AI model will be trained from CFD (computational fluid dynamics) 2-D axisymmetric simulations. To this end, hundreds of stenosis geometries will be simulated with COMSOL using a specific turbulent regime. In this Ph.D. project, the student will first simulate blood flows in axisymmetric geometries to generate a large paired dataset. Once the AI model will be trained and validated with other CFD simulations, the Ph.D. student will perform in vitro experiments in stenotic flows. Geometries and flow rates will be measured with a clinical ultrasound machine. The pressure losses estimated from the AI model will be compared with those measured by pressure wires. Depending on the progress of the project, asymmetrical 3-D geometries will then be addressed.

[1]       Young DF, Tsai FY. Flow characteristics in models of arterial stenoses—II. Unsteady flow. Journal of biomechanics. 1973 Sep 1;6(5):547-5