Uncertainty evaluation from non-invasive estimation of temperature variation using B-mode ultrasonic images from a plastic phantom

• Average grey-level (AVGLAVGL) from B-mode images is assessed for temperature tracking.
• A regression model for temperature/AVGLAVGL relation is presented for 37°C to 46°C.
• A first approach of uncertainty model is proposed for temperature/AVGLAVGL relation.
• Result average ratio of 2.03°C/AVGL‾ and expanded uncertainty of 0.36°C/AVGL‾.
• Leave-one-out test confirms the regression model generalisation capability.

The average grey-level (AVGLAVGL), calculated from B-mode images, has been assessed as a tool for non-invasive temperature tracking in phantoms, and in different sort of tissues. However, the respective uncertainty evaluation, which is an important tool to support measurement results, has not been studied up to now. Herein, a plastic phantom was subjected to 10 consecutive heating tests at repeatability condition, with temperature varying from 37°C to 46°C. At each test, the sample was continuously imaged with an ultrasound scanner, while temperature was measured every 5 s using two calibrated Type-T thermocouples placed at the central region and at the border of the phantom. AVGL   values were calculated from the acquired B-mode images and correlated with temperature variations. Then, a regression model for the temperature/AVGLAVGL relation was obtained, and a first approach of uncertainty model is proposed. The result shows an average ratio of 2.03°C/AVGL‾ and an expanded uncertainty of 0.36°C/AVGL‾. Moreover, a leave-one-out test was performed among the different tests, confirming the regression model generalisation capability.