ABSTRACT
Digitally reconstructed radiographs (DRRs) have played an important role in radiotherapy, especially when it comes to patient alignment for image-guided radiotherapy (IGRT) before treatment delivery. DRR calculations have also received much attention with the advent
CONTENTS 2.1 Introduction 15
2.1.1 Applications and Motivation 16 2.2Radiological Path and Digitally Reconstructed Radiographs 17
2.2.1Beer-Lambert Law 17 2.2.2Integration Problem 18
2.3Siddon’s Algorithm 19 2.3.1Modern Incremental Algorithm 21 2.3.2Siddon’s Incremental Algorithm on GPU 21
2.4Fixed Stride Interpolation 22 2.4.1Fixed Stride Interpolation on GPU 23
2.5Texture Stacking (Shear-Warp) and Texture Slicing 24 2.5.1Texture Stacking (Shear-Warp) on GPU 25 2.5.2Texture Slicing on GPU 25 2.5.3Voxel Splatting on GPU 26
2.6Polyenergetic (Energy-Dependent) DRR 26 2.7 Summary 28 References 28
of clinically viable statistical and iterative cone-beam computed tomography (CBCT) reconstruction techniques, where the computational complexity of the DRR calculation tends to be a bottleneck. e fundamental ray tracing component used in DRRs is also used in advanced Monte Carlo (MC) dose calculations and tends to be a performance bottleneck as well. In this chapter, we will explore the variety of DRR and ray tracing techniques, which have been implemented on GPU, with a focus on the imaging domain. We continue with an overview of relevant applications, then, in the following sections, set up the physical concept and integration problem, and review recently published GPU-based DRR techniques.
