A convolution‐superposition dose calculation engine for GPUs

Abstract

These results suggest that GPUs are an attractive solution for radiation therapy applications and that careful design, taking the GPU architecture into account, is critical in obtaining significant acceleration factors. These results potentially can have a significant impact on complex dose delivery techniques requiring intensive dose calculations such as intensity-modulated radiation therapy (IMRT) and arc therapy. They also are relevant for adaptive radiation therapy where dose results must be obtained rapidly.

A fast three‐dimensional gamma evaluation using a GPU utilizing texture memory for on‐the‐fly interpolations

“…The GPU is a powerful, massively parallel programmable architecture, 7 which can be used to decrease the calculation time for many applications. In radiotherapy, the GPU has already been used in a wide range of applications (e.g., ray-tracing, dose calculation, and CT reconstruction techniques) [8][9][10][11][12][13][14] reporting speed-up ratios from 6 to 908.…”

Monte Carlo Based Ray Tracing in CPU-GPU Heterogeneous Systems and Applications in Radiation Therapy

“…Since the patient treatment space is commonly represented as a uniform grid in dose calculation applications, LE-MCBRT can be directly applied to known MC based dose calculation implementations, such as GPUMCD [13], gDPM [15], and GMC [18]. Other than stochastic MC dose calculation methods, some deterministic approaches (e.g., collapsed cone convolution superposition [14] and pencilbeam [8]) also use ray tracing algorithms to calculate the energy propagation for massive numbers of particles. LE-MCBRT can also help accelerate these approaches.…”

“…The locations of particle-medium interactions (and the amount of transferred energy) are randomly sampled. Second, the deterministic convolution superposition (CS) method [14] is used to spread energy around the interaction points. The speed advantage of MCCS over the MC methods and its accuracy advantage over the CS methods make it a good candidate for constructing a routine dose calculation tool for clinical radiation treatment planning.…”

“…Hence, MCBRT is often considered to be able to faithfully capture the real physical interaction processes [16]. For example, in radiation treatment planning, Monte Carlo (MC) dose calculation methods using MCBRT are viewed as the most accurate approach and used to evaluate the quality of deterministic dose calculation methods [14]. But, achieving high accuracy using MCBRT often requires to process a huge number of randomly generated rays.…”

GPU-based high-performance computing for radiation therapy

“…A speed-up of over 100 over the highly optimized Pinacle (Philips, Madison, WI) implementation was observed on an NVIDIA GTX280 card, where the absolute computation time was about 1 second per plan. Later, Hissoiny developed another implementation of the SC algorithm on GPU (Hissoiny et al , 2010). Special function unit on the GPU was utilized to accelerate the computations of intrinsic functions, e.g.…”