Abstract
2632
Objectives So semi-quantitatitve assessments can be performed when a LV-in-torso phantom is imaged on a multi-pinhole cardiac imager, normal files of 6 phantom configurations were constructed. [The files are specific for the Data Spectrum Cardiac Insert in Anthropomorphic Torso acquired, reconstructed, and analyzed using GE Discovery NM 530c, GE Myovation, and Syntermed Emory Cardiac Toolbox.]
Methods Three distributions in the LV-in-torso phantom were imaged: 22 MBq of Tc-99m in myocardium, none in torso or liver; 22 in myocardium, 185 in torso, none in liver; 22 in myocardium, 185 in torso, 125 in liver. Each distribution was imaged 25 times supine, 25 times prone. Following each acquisition, the phantom was removed from the imager and positioned again in the system. Each acquisition required appropriate positioning of the detectors. Each acquisition was 4 min. Each data set was reconstructed using default parameters (no AC or SC). Six normal files based on 25 studies each were constructed. To demonstrate use of the files, 6 sets of data of the LV with full thickness LAD and half thickness RCA defects were acquired and processed. Activities in the LV-in-torso phantom were as above; each of the 3 distributions was imaged 5 times supine and 5 times prone. Each data set was processed with the appropriate normal file; the 2 defects were assessed individually and collectively as percent of total myocardium. Three coefficients of variation were computed for each of the 6 sets of 5 measurements.
Results The coefficients of variation for the LAD defect percent of total ranged from 0.06 to 0.13; the RCA defect percent of total from 0.11 to 0.27; the collective percent of total from 0.09 to 0.18. The greatest variations were in measurements of the RCA defect--that most affected by attenuation and distance from detectors.
Conclusions Normal files for a left-ventricle phantom imaged with a multi-pinhole cardiac imager can be generated and used in characterization studies of the novel imager.