Side Lobe Reduction Using Centroid and Flatness in Passive Cavitation Imaging

Abstract

Cavitation bubbles generated by high-intensity focused ultrasound strongly collapse and emit ultrasound waves. Since passive cavitation imaging can be performed using only receive focusing, side lobes appear over a wide area of an image. The delay and sum beamforming technique used to construct passive cavitation images cannot effectively image the cavitation occurrence time and spatial distribution because a small cavitation signal is buried in the side lobe of a large cavitation signal when there are a large number of cavitation bubbles. The ultrasound wave emitted when the cavitation bubble collapses is a short duration pulse. After compensating for the focusing delay, the signals due to the main lobe (at the imaging point) received at each element of the transducer array are similar in magnitude, but those due to the side lobes (other than the imaging point) is significant only at some elements of the array. Using this characteristic of the received channel data, we propose both centroid and flatness as metrics to evaluate the effect of the main and side lobes at imaging points. If the centroid of the signal magnitude distribution of the entire receiving channel is positioned at the center of the array, the signal is considered to be due to the main lobe. However, if the centroid is computed to be located near both ends of the array, the signal is considered to be due to the side lobes. If the signals are found to be due to the side lobes in passive cavitation imaging, the pixel brightness is reduced by weighting using the centroid and flatness metrics. We computed the centroid and flatness metrics using computer simulation and experimental data, and confirmed the suppression of side lobes by multiplying the image by a weight using the centroid and flatness. The proposed method makes it easier to observe cavitation by effectively removing side lobes from passive cavitation images when the number of cavitation bubbles is not large.