Graphics Processing Unit-Based Bioheat Simulation to Facilitate Rapid Decision Making Associated with Cryosurgery Training.
Graphics Processing Unit-Based Bioheat Simulation to Facilitate Rapid Decision Making Associated with Cryosurgery Training.
复制标题
DOI：
10.1177/1533034615580694
复制
发表时间：
2016-04
影响因子：
2.8
通讯作者：
Rabin Y
中科院分区：
文献类型：
--
作者：
Keelan R;Zhang H;Shimada K;Rabin Y
这项研究的重点是在图形处理器(GPU)上实现一种高效的冷冻外科模拟数值技术，作为加速运行时间的替代手段。这项研究是为冷冻外科开发计算机化培训工具的持续努力的一部分，以前列腺冷冻外科为发展模式。快速模拟各种测试用例的能力对于促进与医疗培训相关的合理决策至关重要。该训练工具与临床实践相一致，旨在将冰冻区域轮廓和相应的温度场与目标区域形状进行关联。目前的研究集中在使用C++加速大规模并行(AMP)进行基于GPU的计算的可行性，作为一种可能的实现。与在多核中央处理器(CPU)上更常见的实施相比，在各种计算平台上的基准结果显示冷冻外科模拟的三倍加速(笔记本电脑)到十三倍加速(游戏机)之间。虽然基于GPU的模拟的一般概念并不新鲜，但它在相变问题中的应用，以及对冷冻外科优化的独特要求，代表了本研究的核心贡献。
This study focuses on the implementation of an efficient numerical technique for cryosurgery simulations on a graphics processing unit (GPU) as an alternative means to accelerate run time. This study is part of an ongoing effort to develop computerized training tools for cryosurgery, with prostate cryosurgery as a developmental model. The ability to perform rapid simulations of varying test cases is critical to facilitate sound decision making associated with medical training. Consistent with clinical practice, the training tool aims at correlating the frozen region contour and the corresponding temperature field with the target region shape. The current study focuses on the feasibility of GPU-based computation using C++ accelerated massive parallelism (AMP), as one possible implementation. Benchmark results on a variety of computation platforms display between three-fold acceleration (laptop) and thirteen-fold acceleration (gaming computer) of cryosurgery simulation, in comparison with the more common implementation on a multi-core central processing unit (CPU). While the general concept of GPU-based simulations is not new, its application to phase-change problems, combined with the unique requirements for cryosurgery optimization, represents the core contribution of the current study.