Frequency domain (f-k) migration methods for 3D ultrasound imaging with matrix arrays

Robert, Sébastien; CEA-LIST; France

Merabet, L.; Institut Langevin, ESPCI Paris; France
Prada, C.; Institut Langevin, ESPCI Paris; France
Robert, S.; CEA, LIST; France

ID: ECNDT-0381-2018
Session: PAUT-Signal processing
Room: G2
Date: 2018-06-13
Time: 15:40 - 16:00

3D ultrasound imaging with matrix arrays is an ideal solution to characterize defects exhibiting complex geometries, or to image a large volume of the material at a fixed position. However, the formation of 3D images is often costly in computation times because of the large number of voxels (up to 107) and array elements (up to 256) to deal with. In order to overcome this problem, we propose to associate frequency-domain (f-k) reconstruction algorithms with plane wave emissions. Compared to the time-domain imaging (such as classical TFM), the Fourier-domain algorithms can theoretically reduce the complexity by a factor up to the number of elements, which is ideally suited for imaging with matrix array probes. In this communication, 3D images are given in realistic NDT configurations (probes immersed in water or equipped with Plexiglas wedges) and it is demonstrated that the reconstruction algorithms in the f-k domain are able to form images much larger than the array aperture. Then, the Fourier-domain imaging methods are compared with the time-domain imaging in terms of computation time, lateral resolution and signal/noise ratio. It is shown that the f-k methods provide 3D images of similar or even better quality according to these criteria, while significantly reducing the reconstruction times.