Real‐time Tissue Thermometry Using an Acoustic Neural Network Method during HIFU Treatment

A recurrent neural network solution was proposed to map changes in multiple acoustic parameters, such as apparent displacement, echo backscattered intensity change, echo signal correlation coefficient and tissue elasticity, to temperature rise during High Intensity Focused Ultrasound (HIFU) dosing. The training and testing of the neural network was based on data from experiments on HIFU treated ex-vivo porcine and bovine liver. The experiments included acquiring 3D echo signals along with directly measuring temperature with thermocouples in the region of the HIFU focal zone (treatment target) before and during the HIFU dosing event. The results demonstrate that the proposed method may provide temperature rise estimation (thermometry) over practical temperature ranges for the therapy, and may reveal the individual acoustic parameters' correlations with temperature rise.