Improved temperature imaging of focused ultrasound thermal therapy using a sigmoid model based cross-correlation algorithm

Focused ultrasound (FUS) thermal therapy relies on temperature imaging for treatment monitoring and guidance. Current ultrasonic temperature estimation mainly involves the detection of echo-time shifts using cross-correlation processing between segmented radio-frequency (RF) data pre- and post-FUS heating. Generally, a minimum RF data segment of around 6 wavelengths is required to obtain accurate echo-time shift estimates and mitigate artifacts. Shorter data segments are essential to retrieve better axial resolution in temperature image; however, shorter data segments generally lead to increased estimation errors. In this work, we propose a sigmoid model based cross-correlation algorithm to address this problem. First, cross-correlation processing with a large non-overlapping window (e.g., around 10 wavelengths) is applied to obtain coarse echo-time shift estimates between RF data pre- and post-FUS heating. According to the feature of the echo-time shift from FUS heating, a sigmoid model is adopted to fit the coarse estimates, and used as the initial estimates for the further cross-correlation processing with a smaller window (e.g., 2 wavelengths) which improves axial resolution and suppress artifacts while providing comparable noise properties to those obtained by the conventional method with a 6-wavelength window. The experimental results showed that this new method can provide accurate echo-time shift estimation and mitigate the artifacts while only requiring less than a 2-wavelength data segment. This work helps to improve axial resolution of the current ultrasonic temperature estimation used for guidance of the FUS thermal ablation procedure.