Thermal strain imaging (TSI) is a promising technique for ultrasonic thermometry, especially in the applications of thermal therapies. The accuracy of TSI is dependent on the sampling rate and line density of B-Scan images, and the prevalent IQ-demodulated ultrasound data outputted from low- and middle-end machines are therefore insufficient. Here, the feasibility of using interpolated IQ images for TSI (based on the "infinitesimal echo strain filter" model) is studied through in vivo experiments targeting the perirenal fat of pigs. It is demonstrated that, axial interpolations, especially those using the zero-padding algorithm, can recover the capabilities of the low-sampling-rate complex IQ images in TSI, and make their performances comparable to those of RF/IQ complex images with higher sample rate. Meanwhile, interpolations along the lateral direction can increase the line density of IQ images, reduce TSI errors, and reveal more details in the temperature maps. In the experiments, the variation in the thermometry coefficient (the k-value) is well below 3%. The findings here bring down the requirement of high sampling rate as well as high line density of US images in TSI, making it possible to be applied on common US machines.
Keywords: Data Interpolation; Thermal strain imaging; Ultrasonic thermometry.
Copyright © 2020 Elsevier B.V. All rights reserved.