Premium
Performance of a novel collimator for high‐sensitivity brain SPECT
Author(s) -
Fakhri Georges El,
Ouyang Jinsong,
Zimmerman Robert E.,
Fischman Alan J.,
Kijewski Marie Foley
Publication year - 2006
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 - 10.1118/1.2143140
Subject(s) - collimator , imaging phantom , scanner , single photon emission computed tomography , nuclear medicine , physics , partial volume , image resolution , correction for attenuation , filter (signal processing) , optics , point spread function , medical imaging , emission computed tomography , radius , computer science , attenuation , positron emission tomography , artificial intelligence , computer vision , medicine , computer security
We assessed improvements in performance in detection and estimation tasks due to a novel brain single photon computed tomography collimator. Data were acquired on theCeraSPECT TMscanner using both new and standard collimators. The new variable focusing collimatorSensOgrade TMsamples the projections unequally, with central regions more heavily represented, to compensate for attenuation of counts from central brain structures. Furthermore, it utilizes more of the cylindrical crystal surface. Two phantom studies were performed. The first phantom was a 21‐cm‐diameter cylindrical background containing nine spheres ranging from 0.5 to 5cm 3in volume.Tc99 msphere to background activity ratio was 10:1. Twenty‐nine 10‐min datasets were acquired with each collimator. The second phantom was the Radiology Support Devices (Long Beach, CA) striatal phantom with striatal‐background ratios of 10:1 on the left and 5:1 on the right. Twenty‐nine 4‐min datasets were acquired with each collimator. Perfusion imaging usingTc99 m‐HMPAO was also performed in three healthy volunteers using both collimators under identical simulations. Projections were reconstructed by filtered backprojection with an unwindowed ramp filter. The nonprewhitening matched filter signal‐to‐noise ratio (NPW‐SNR) was computed as a surrogate for human performance in detecting spherical lesions. Sphere activity concentration, radius, and location coordinates were simultaneously estimated by fitting images to an assumed model using an iterative nonlinear algorithm. Resolution recovery was implicit in the estimation procedure, as the point spread function was incorporated into the model. NPW‐SNR for sphere detection was 1.5 to 2 times greater with the new collimator; for the striatal phantom the improvement in SNR was 54%. The SNR for estimating sphere activity concentration improved by 46 to 89 % for spheres located more than 5 cm from the phantom center. Images acquired with the standard collimator were too noisy in the central regions to allow estimation of sphere activity. InTc99 m‐HMPAO human studies, SNR was improved by 21 to 41 % in the cortex, 66% in the basal ganglia, and 74% in the thalamus. The new collimator leads to substantially improved detection and estimation performance throughout the brain. The higher sensitivity will be particularly important for dynamic imaging.