Is the Average Gray-Level from Ultrasound B-Mode Images Able to Estimate Temperature Variations in Ex-Vivo Tissue?

摘要

This paper presents the first results of a method to estimate temperature change by monitoring the gray-scale average value of standard ultrasound (US) images. It was carried out an experiment with a bovine muscle sample with four thermocouples along its depth, spaced of 1 cm. The sample was immersed in a reservoir to which water at 50°C was added. Images were recorded with a commercial ultrasound equipment during sample heating and cooling (2.5 h experiment). Temperature was acquired every 10 s. Each image was then divided in four horizontal regions-of-interest (ROIs), corresponding to each thermocouple positioning. Average gray-scale values versus temperature change were obtained and have shown a clear linear pattern within the range 0.5 to 8°C for all ROIs. A linear model was proposed to fit the curves with a maximum prediction error of approximately 0.6°C, which is close to the accepted error for hyperthermia treatment (0.5°C). The next step is to use ultrasound also as the heating source.