This research project is related to high-frame-rate 3D ultrasound cardiac imaging. Provide clinicians with high-quality (contrast / resolution) 3D images at high frame rates (> several hundreds frames per second) remains a real challenge. As a consequence, this topic is one of the priorities in the ultrasound community, both for the academic and industrial groups.
The method we intend to develop is based on diverging wave (DW) imaging. In DW imaging, an ultrasound field is first produced with all the elements of a matrix array to reproduce the acoustic field that would originate from a virtual source located behind the aperture. Unfortunately, it is not sufficient to collect the echoes from one single insonification; several transmissions with different virtual sources need to be compounded. Even if this acquisition scheme is extremely fast, phase changes due to cardiac motion can occur and produce destructive interferences in the raw signals. This prevents optimal alignment of the echoes and reduces the final image quality. Motion compensation before coherent compounding can improve this process in 3D, as has been demonstrated in 2D imaging.
Our preliminary results (Joos et al, IEEE IUS 2017) have shown that compensating the axial motion obtained from tissue Doppler is a promising approach. But the locations of the virtual sources and their arrangement and ordering still need to be optimized. For this task, we might use approaches such as those used for sparse array optimization (Roux et al., IEEE IUS 2017), or some techniques inspired from deep learning (Gasse et al., IEEE Trans. UFFC 2017).
The 3D motion compensation technique (3D MOCO) will first be tested in simulations and will then be validated with phantoms and a 1024-channel system driving a 32x32 matrix array. A dynamic heart phantom, as well as a vortex flow phantom with known ground-truth, will be used.
Attractiveness of Lyon’s environment :
To implement such sequences, the access to a high-density channel system is needed. Less than five places in the world are equipped with such equipment. Owing to a collaboration between two ultrasound research units (Labtau and Creatis; Petrusca et al., IEEE IUS 2017), Lyon area is one of these places.
The ultrasound team at CREATIS is among the first to start working on 3D MOCO (Joos et al., IEEE IUS 2017). Damien Garcia, now researcher at Creatis, and involved in this Post-Doc project, paved the way for 2D MOCO (Porée et al., IEEE TMI 2017).
Hervé Liebgott, PU UCBL, firstname.lastname@example.org