OBJECTIVE: To determine if computer registration improves the image fusion of hybrid positron emission tomography (PET)/computed tomography (CT) data sets in the thorax acquired during different breathing maneuvers. METHODS: Hybrid PET/CT scans were acquired with varying CT respiratory instructions. The scans of 64 patients with 5 different breathing maneuvers were reviewed, including expiration, suspended breath hold, quiet breathing, small breath in, and regular breath in. The PET/CT data sets were reviewed before registration, after linear registration, and combined linear/nonlinear registration. Each PET/CT data set was graded for the quality of alignment at 5 anatomical locations (diaphragm, aortic arch, heart, thoracic spine, and lung apices) and rated from 1 (very poor) to 5 (excellent). The Kruskal-Wallis test was used to compare alignment between the breathing protocols, and Wilcoxon signed rank test was used to compare the registration techniques. RESULTS: The quality of anatomical alignment between superimposed (nonregistered) PET and attenuation correction CT (CTAC) acquired during expiration, suspended breath hold, and quiet breathing was excellent, with no significant changes in alignment following registration. The quality of anatomical alignment between superimposed (nonregistered) PET and CTAC acquired during inspiration showed significant misalignment at the heart (P = 0.001, P < 0.0001) and diaphragm (P = 0.0001, P < 0.0001) which did not correct with linear registration. Alignment significantly improved with additional nonlinear registration at the diaphragm (P = 0.008, P = 0.0002), although cardiac misalignment remained. CONCLUSIONS: With hybrid PET/CT, excellent anatomical alignment between PET and CTAC scans can be achieved with CT acquisition during expiration, suspended breath hold, or quiet breathing, and additional software registration is not necessary. Nonlinear registration significantly improves alignment discrepancies at the diaphragm if inspiratory methods are used, however, cardiac misalignment remains.
OBJECTIVE: To determine if computer registration improves the image fusion of hybrid positron emission tomography (PET)/computed tomography (CT) data sets in the thorax acquired during different breathing maneuvers. METHODS: Hybrid PET/CT scans were acquired with varying CT respiratory instructions. The scans of 64 patients with 5 different breathing maneuvers were reviewed, including expiration, suspended breath hold, quiet breathing, small breath in, and regular breath in. The PET/CT data sets were reviewed before registration, after linear registration, and combined linear/nonlinear registration. Each PET/CT data set was graded for the quality of alignment at 5 anatomical locations (diaphragm, aortic arch, heart, thoracic spine, and lung apices) and rated from 1 (very poor) to 5 (excellent). The Kruskal-Wallis test was used to compare alignment between the breathing protocols, and Wilcoxon signed rank test was used to compare the registration techniques. RESULTS: The quality of anatomical alignment between superimposed (nonregistered) PET and attenuation correction CT (CTAC) acquired during expiration, suspended breath hold, and quiet breathing was excellent, with no significant changes in alignment following registration. The quality of anatomical alignment between superimposed (nonregistered) PET and CTAC acquired during inspiration showed significant misalignment at the heart (P = 0.001, P < 0.0001) and diaphragm (P = 0.0001, P < 0.0001) which did not correct with linear registration. Alignment significantly improved with additional nonlinear registration at the diaphragm (P = 0.008, P = 0.0002), although cardiac misalignment remained. CONCLUSIONS: With hybrid PET/CT, excellent anatomical alignment between PET and CTAC scans can be achieved with CT acquisition during expiration, suspended breath hold, or quiet breathing, and additional software registration is not necessary. Nonlinear registration significantly improves alignment discrepancies at the diaphragm if inspiratory methods are used, however, cardiac misalignment remains.
Authors: Elizabeth A Smith; Bertrand P Cinquin; Gerry McDermott; Mark A Le Gros; Dilworth Y Parkinson; Hong Tae Kim; Carolyn A Larabell Journal: J Struct Biol Date: 2013-03-24 Impact factor: 2.867