French
English
Thesis Abstract :
4D Flow MRI is a phase contrast imaging technique that allows a comprehensive assessment of cardiovascular function by providing volumetric measurements of blood flow throughout the cardiac cycle. The phase of the MRI signal is proportional to velocity and is limited by the velocity encoding parameter (VENC), which restricts the velocity range to -VENC and +VENC. 4D flow MRI can capture complex blood flow patterns, including low velocities in veins and high velocities in arteries. Choosing the right VENC is crucial: a low VENC causes aliasing when velocities exceed ±VENC, while a high VENC introduces noise, making slow flow measurements inaccurate. The optimum VENC balances a good velocity-to-noise ratio (VNR) for accurate slow-flow measurements with a wide dynamic range to avoid aliasing. To extend the velocity dynamic range, dual or multiple VENC techniques are used, acquiring data with different VENC values. Standard dual-VENC unwrapping uses the VENC_high data to detect phase jumps in the VENC_low data, adding or subtracting multiples of 2π to combine the benefits of both acquisitions. However, in pathological cases, this method is limited by aliasing in the VENC_high data due to unexpectedly high velocities. To address this issue, we have introduced a new 4D Flow MRI double-VENC sequence, based on the coprime rule for VENC ratio, allowing a wide dynamic range of velocities, accompanied by a time-efficient velocity unwrapping algorithm, validated in vitro and demonstrated in vivo in patients with cardiovascular pathologies. Despite these advances, double-VENC sequences are limited by long acquisition times. Radial 3D sampling has emerged as a promising solution, preserving essential low-frequency data during undersampling, and being more resistant to motion artefacts. Using free-flow acquisitions and auto-gating techniques, 4D flow MRI with 3D radial sampling allows cardiac and respiratory signals to be extracted directly from k-space data, eliminating the need for external devices such as ECGs. We also investigated the performance of the dual-VENC coprime sequence combined with 3D radial sampling to overcome the acquisition time limitations of dual-VENC methods. Although 4D flow MRI offers detailed measurements, its time-consuming nature and high cost make it less practical compared to echocardiography, particularly colour Doppler, which is affordable, portable and offers real-time imaging. However, the one-dimensional nature of colour Doppler and its dependence on angle of incidence limit its ability to capture complex three-dimensional flow patterns. Techniques such as intraventricular vector flow mapping (iVFM) have been developed to extract two-dimensional velocity fields from colour Doppler data, providing a more accurate representation of blood flow dynamics. Although the iVFM has been validated by computational fluid dynamics (CFD) simulations, challenges remain when comparing its results with those of 4D flow MRI, the standard for in vivo blood flow velocity measurements. A major challenge is the possibility of discrepancies when comparing instantaneous velocity fields derived from iVFM with time-averaged data from 4D flow MRI. In this thesis, we developed a methodology to reconcile these discrepancies by comparing the velocity fields measured by the two techniques within the left ventricle.
The PhD Jury will be composed by :
Dyverfeldt, Petter Professor, Linköping University, Sweden, Rapporteur
Celi Simona, Adjoint Professor, University of Pisa, Italy, Rapporteure
Ratiney Hélène, Directrice de recherche, CNRS, CREATIS Lyon, Examinatrice
Soulat Gilles, PU-PH, Radiologue, Université Paris Cité médecine, Examinateur
Sigovan Monica, Chargée de recherche, CNRS, CREATIS Lyon, Directrice de thèse
Garcia Damien, Directeur de recherche, INSERM, CREATIS Lyon, Co-directeur de thèse
Boccalini Sara, Radiologue, Hospices Civils de Lyon, Invitée