We are working on improving the spatial resolution of ultrasound images for medical ultrasound imaging and non-destructive testing. (1) Optimization techniques such as unique regularization, maximum a posteriori, and Wiener filters have been employed in inverse filtering. The distribution of reflectivity and the backscattering coefficient can be estimated quantitatively. The point spread function is estimated based on the local autocorrelation function. In addition to conventional scanning with a focused beam, we perform classical synthetic aperture processing and reception dynamic focusing with plane wave transmission. For focused beam scanning, stationarity is assumed at every depth or approximately 5 mm, whereas for other cases, across larger regions, processing is performed in the spatial, or the faster frequency domain. (2) Harmonic generation through phase matching for pulse inversion or in silico nonlinear processing is also effective. Calculating the power of the signal itself before or after beamforming using the received signal, or multiplying waves of different frequencies or steering angles, makes it possible to achieve high frequency, wide bandwidth and detection simultaneously. Combinations of (1) and (2) and coherent compounding have also been reported. Deep learning and phase are also employed. These increase the precision of structure positioning considerably. Details will be reported in POMA, etc.
Chikayoshi Sumi (Wed,) studied this question.