Quantitative Measures for Evaluation of Ultrasound Therapies of the Prostate

Abstract

Development of non‐invasive techniques for prostate cancer treatment requires implementation of quantitative measures for evaluation of the treatment results. In this paper. we introduce measures that estimate spatial targeting accuracy and potential thermal damage to the structures surrounding the prostate. The measures were developed for the technique of treating prostate cancer with a transurethral ultrasound heating applicators guided by active MR temperature feedback. Variations of ultrasound element length and related MR imaging parameters such as MR slice thickness and update time were investigated by performing numerical simulations of the treatment on a database of ten patient prostate geometries segmented from clinical MR images. Susceptibility of each parameter configuration to uncertainty in MR temperature measurements was studied by adding noise to the temperature measurements. Gaussian noise with zero mean and standard deviation of 0, 1, 3 and 5° C was used to model different levels of uncertainty in MR temperature measurements. Results of simulations for each parameter configuration were averaged over the database of the ten prostate patient geometries studied. Results have shown that for update time of 5 seconds both 3‐ and 5‐mm elements achieve appropriate performance for temperature uncertainty up to 3° C, while temperature uncertainty of 5° C leads to noticeable reduction in spatial accuracy and increased risk of damaging rectal wall. Ten‐mm elements lacked spatial accuracy and had higher risk of damaging rectal wall compared to 3‐ and 5‐mm elements, but were less sensitive to the level of temperature uncertainty. The effect of changing update time was studied for 5‐mm elements. Simulations showed that update time had minor effects on all aspects of treatment for temperature uncertainty of 0° C and 1° C, while temperature uncertainties of 3° C and 5° C led to reduced spatial accuracy, increased potential damage to the rectal wall, and longer treatment times for update time above 5 seconds. Overall evaluation of results suggested that 5‐mm elements showed best performance under physically reachable MR imaging parameters.