Goulven LE-MOIGN : Design and optimisation of a random 2D sparse array for 3D medical ultrasound imaging

Abstract :

3D ultrasound imaging is one of the top priorities in medical imaging. There are several ultrasound imaging applications: from conventional echography to cardiac strain estimation, 3D expansion makes it possible to refine the results making the diagnosis more relevant for the clinician. In conventional ultrasound, a linear probe composed of elements - often piezoelectric - allows the transmission and the reception of ultrasound signals. This enables the reconstruction of images. In 3D imaging, a 2D probe is required to avoid mechanical scanning. If we want to conserve a good image quality, using the aforementioned method, we need to strongly increases the number of transducers.

In order to limit computation time and connectivity, we need to keep number of elements as low as possible. Therefore, we consider a sparse distribution of said elements. Because regular distribution of the elements makes artifacts on the images, we need to break that regularity and choose an uneven distribution. This can be done via a deterministic or random way; the second solution is elected but need an optimisation step.

The first part of the thesis focuses on the optimisation of a random sparse 256 elements probe, transmitting diverging waves, for fast ultrasound cardiography. A second part will be devoted to the implementation of an imaging algorithm. At the end, we should be able to build our own probe thanks to the collaboration with the University of Sherbrooke (co-tutelle).