Premium
SU‐G‐IeP2‐06: Evaluation of Registration Accuracy for Cone‐Beam CT Reconstruction Techniques
Author(s) -
Li J,
Zhang H,
Wang P
Publication year - 2016
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4957011
Subject(s) - image registration , cone beam computed tomography , cone beam ct , nuclear medicine , medicine , image guided radiation therapy , head and neck , radiation therapy , computed tomography , artificial intelligence , computer science , computer vision , radiology , image (mathematics) , surgery
Purpose: Cone‐beam (CB) computed tomography (CT) is used for image guidance during radiotherapy treatment delivery. Conventional Feldkamp and compressed sensing (CS) based CBCT recon‐struction techniques are compared for image registration. This study is to evaluate the image registration accuracy of conventional and CS CBCT for head‐and‐neck (HN) patients. Methods: Ten HN patients with oropharyngeal tumors were retrospectively selected. Each HN patient had one planning CT (CTP) and three CBCTs were acquired during an adaptive radiotherapy proto‐col. Each CBCT was reconstructed by both the conventional (CBCTCON) and compressed sens‐ing (CBCTCS) methods. Two oncologists manually labeled 23 landmarks of normal tissue and implanted gold markers on both the CTP and CBCTCON. Subsequently, landmarks on CTp were propagated to CBCTs, using a b‐spline‐based deformable image registration (DIR) and rigid registration (RR). The errors of these registration methods between two CBCT methods were calcu‐lated. Results: For DIR, the mean distance between the propagated and the labeled landmarks was 2.8 mm ± 0.52 for CBCTCS, and 3.5 mm ± 0.75 for CBCTCON. For RR, the mean distance between the propagated and the labeled landmarks was 6.8 mm ± 0.92 for CBCTCS, and 8.7 mm ± 0.95 CBCTCON. Conclusion: This study has demonstrated that CS CBCT is more accurate than conventional CBCT in image registration by both rigid and non‐rigid methods. It is potentially suggested that CS CBCT is an improved image modality for image guided adaptive applications.