ABSTRACT
The γ-index concept introduced by Low et al. [1] has been widely used to compare two dose distributions in cancer radiotherapy. With the recently developed and more sophisticated treatment modalities such as volumetric-modulated arc therapy (VMAT) [2], the comparison of
CONTENTS 20.1 Introduction 333 20.2 Graphic Interpretations of Gamma Index and Algorithm 335
20.2.1Original Gamma Index Algorithm 335 20.2.2Geometric Interpretation of Gamma Index 336 20.2.3Modied Gamma Index Algorithm 338
20.3GPU Implementations and Optimization of Performance 339 20.4Experimental Results 341
20.4.1Evaluation Data Set and Time Measurement Criteria 341 20.4.2Speedup of GPU versus CPU 341 20.4.3Eect of DD Sorting on Computation Time T Gg 343 20.4.4 Eect of γ-Index Values on Computation Time T Cg and T Gg 343 20.4.5 Eect of Dose Distribution Resolution on the
Computation Time T Cg and T Gg 345 20.4.6 Eect of DD and DTA Criteria on Computation Time
T Cg and T Gg 345 20.5 Conclusion and Discussions 347 References 348
three-dimensional (3D) dose distributions becomes necessary in patientspecic treatment plans quality assurance (QA). However, comparing two 3D dose distributions is still a time-consuming (e.g., many minutes) task, because of the involvement of computational intensive tasks such as interpolation of dose grid and exhaustive search. ere is a clinical need to signicantly speed up 3D γ-index computations.
