Ultrasound & Heat Transfer Modeling
Treatment planning for high-intensity focused ultrasound requires repeated iteration of a model in order to optimize the treatment parameters. In order for traditional modeling techniques to have enough accuracy to be meaningful, they can have a huge computational cost. Reduced order modeling is a data-driven modeling technique that aims to reduce the computational cost of a model by identifying and exploiting underlying patterns in the data while still maintaining enough accuracy to be meaningful, thus effectively expediting treatment plan optimization.
Modeling where ultrasound goes in a human body can be tricky. Different types of tissue have different sound speeds and attenuation, which means that the path and intensity of the pressure wave is constantly changing. To model this accurately, we use the hybrid angular spectrum method.
We are currently working on a computational bioheat transfer model, which predicts the heating in tissues given a pressure pattern. The goal is to modify an existing explicit solver to become an implicit-type solver.
Transcranial magnetic resonance-guided focused ultrasound (MRgFUS) surgery is a noninvasive procedure used to treat movement disorders like essential tremor. When MRgFUS is used in the brain, the ultrasound is absorbed by all surrounding tissue, especially the skull. When target tissues are central in the brain, ultrasonic waves are widely distributed around the skull, and concerns of excessive heating at the skull-brain interface is small. However, targeting tissue closer to the skull leads to a more concentrated application of ultrasound. This increases the risk of damaging healthy brain tissues. In current transcranial applications of MRgFUS, cold water is run over the scalp between sonications to prevent skin burns. The purpose of our sensitivity study is to evaluate whether this same external cooling method could reduce heating at the skull-brain interface and thus facilitate MRgFUS treatment of tissues closer to the skull.
