A two-parameter method for the estimation of ultrasound-induced temperature
artifacts was evaluated and compared with other commonly applied methods u
sing analytical solutions to the bioheat equation. The two parameters are t
he exponent of the assumed temperature decay curve after power is turned of
f and the baseline temperature. These parameters are found by optimizing th
e fit of the temperature data from 30 to 60 s after power is turned off. Th
e artifact is modelled as a point source at the centre of a Gaussian temper
ature distribution. The blood flow, baseline temperature, and variance of t
he Gaussian temperature distribution were varied to simulate different clin
ical situations. Noise was added to the model to investigate the effects of
thermometry resolution and sampling intervals. It was found that for artif
acts of < 2 degrees C the two-parameter method had errors of less than 0.25
degrees C, whereas other methods generally had greater errors depending on
the conduction rate and blood flow rate. The effects of the temperature sa
mpling interval and resolution on the ability of the methods to estimate th
e artifact were also investigated, and it was found that the two-parameter
method was much more sensitive to these parameters than other commonly appl
ied methods.