Comparison of Thermal Dose Models in Canine Brain Using the Dice Similarity Coefficient

Joshua Yung¹, Anil Shetty¹, Adam C. Springer¹, Jeffrey Weinberg², John D. Hazle¹, Roger Jason Stafford¹

Various thermal dose models exist for predicting thermal damage from the temperature history provided by MRTI. However, within each model, different parameters or thresholds are often employed. In this work, the performance of multiple thermal dose models were compared using the Dice similarity coefficient (DSC) to determine the spatial overlap between predicted damage for in vivo canine brain during thermal therapy and post-treatment imaging. In all subjects, previously reported thresholds for each model resulted in a high DSC, but were improved by changes in thresholds, which could lead to a more accurate prediction of tissue response.