Towards Exascale Parallel Mesh Generation and Real-Time Medical Image Computing

22 Jun 2016
15:30-16:00
XENIA HOTEL, PORTARIA

Towards Exascale Parallel Mesh Generation and Real-Time Medical Image Computing

In this talk we will present a new approach for guaranteed quality extreme-scale parallel mesh generation and its broader impact to real-time medical image computing. First, we will present a telescopic algorithmic approach to explore a billion-way concurrency by using a hierarchy of abstractions that leverage the memory and network hierarchy of current and emerging multi-layered parallel architectures. Then we will present our work on a parallel runtime software system for implementing effectively mesh generation codes with a billion-way concurrency.

In the second part of this talk we will present a real-time, non-rigid registration method for aligning pre-operative MRI with intra-operative MRI to compensate for brain deformation during tumor resection and Deep Brain Stimulation (DBS). The new method leverages our work on parallel mesh generation and it reduces the alignment error up to seven and six times compared to a widely used rigid and ITK’s FE-based non-rigid registration methods, respectively. Finally, we will present preliminary results on the anatomic modeling for: (i) blood flow diversion of Cerebral Aneurysms and (ii) surgical simulation of Arteriovenous Malformations (AVM).