This PFI: AIR Technology Translation project focuses on translating high performance mathematical models of bioheat transfer for reliably and accurately predicting and visualizing the outcome of laser induced thermal therapy. Minimally-invasive laser ablation is a medical procedure that provides a means of rapidly delivering heat to target diseased tissue in the body and will be used in this project to kill focal cancerous lesions in brain as well as diseased neurological tissue, such as epilepsy. The translated computing technology has the following unique features: (1) the predictive capabilities of the prototype device will assist in minimizing the surgical impact on the patient and (2) the prototype device will be tightly coupled to existing FDA approved procedures in humans and rigorously validated to assure accurate predictions. This provides exemplary improvement in the efficacy of the procedure as no comparable technology currently exists and the neurosurgeon does not have the capability to a-priori visualize outcomes for complex treatment scenarios (multiple lasers/trajectories) near essential anatomical structures.
The project accomplishes this goal by utilizing hybrid multi-core and GPU computing architectures combined with sophisticated mathematical algorithms resulting in a portable, aggressively parallel, medical image driven prototype simulation device. The partnership engages industry (BioTex Inc.) and academic centers (Rice University and MD Anderson) to provide guidance in this minimally invasive neurosurgical market space as well as to commercialize and validate the technology as they pertain to the potential to translate the high performance computing technology along a path that may result in a competitive commercial reality. The potential economic impact is expected to improve treatment effectiveness within the yearly greater than 200,000 brain tumor cases and greater than 1 million epilepsy cases in the U.S. Within a 5-yr timeframe, this will contribute to the U.S. competitiveness in this minimally invasive neurosurgical market space. The societal impact, long term, will provide novel computational and mathematical tools for improving the safety and efficacy of computer assisted image guided therapy for this important medical application.
Magnetic resonance guided laser induced thermal therapy (MRgLITT) is a novel minimally invasive technique used in treating brain tumor patients with no remaining conventional treatment options. Research within this project has focused on developing computer models for treatment planning of MRgLITT in diseased neurological tissue. A computer model that can accurately and reliably predict the outcome of the therapy is a very powerful technology that may be used in planning the procedure to deliver a safe and effective therapy. In particular, we have developed computer models for predicting the outcome of laser-induced ablation of brain tumors. Similar to how a neurosurgeon utilizes previous experiences acquired during his training, we have implemented computer models that utilize imaging information acquired during previous patient MRgLITT therapies to predict the therapy outcome. Current implementations achieve 70% accuracy in predicting the therapeutic kill zone of the tumor. Further, the high performance algorithms for predicting the bioheat transfer have been implemented on graphics processing co-processor technology that has enabled an operating room portable hardware solution. The system interacts with three dimensional visualizations of medical imaging data and the run time of the simulations is well within clinical time constraints. Software used in implementation of the technology is provided open-source through github. http://github.com/ImageGuidedTherapyLab http://tcew.github.io/#/Software
