Diagnostic impact of SPECT image display on assessment of obstructive coronary artery disease.

Published

Journal Article

BACKGROUND: Diagnostic assessment of myocardial perfusion impacts the management of patients with suspected coronary artery disease (CAD). Although various image displays are available for single photon emission computed tomography (SPECT) interpretation, the effects of display differences on SPECT interpretation remain undetermined. METHODS AND RESULTS: We studied 183 patients undergoing SPECT, including 131 consecutive patients referred for angiography and 52 at low CAD risk. Studies were visually interpreted by use of color and gray images, with readers blinded to the results of the other display. In accordance with established criteria, a summed stress score (SSS) of 4 or greater was considered abnormal. The prevalence of abnormal SPECT findings was higher with gray images than with color images (54% vs 48%, P < .001) based on a uniform criterion (SSS > or =4). However, color images yielded equivalent sensitivity (79% vs 82%, P = .7) and improved specificity for global (50% vs 33%, P = .02) and vessel-specific CAD involving the right coronary artery (P < .01) and left anterior descending artery (P < .05). When the criterion for gray images was adjusted upward (SSS > or =5) to reflect increased mean defect severity (SSS of 5.1 vs 4.4, P = .01), gray and color images provided equivalent sensitivity and specificity for global and vessel-specific CAD. CONCLUSIONS: SPECT interpretation can vary according to image display as a result of differences in perfusion defect severity. Adjustment of abnormality criteria for gray images to reflect minor increases in defect severity provides equivalent diagnostic performance of gray and color displays for CAD assessment.

Full Text

Duke Authors

Cited Authors

  • Weinsaft, JW; Gade, CL; Wong, FJ; Kim, HW; Min, JK; Manoushagian, SJ; Okin, PM; Szulc, M

Published Date

  • September 2007

Published In

Volume / Issue

  • 14 / 5

Start / End Page

  • 659 - 668

PubMed ID

  • 17826319

Pubmed Central ID

  • 17826319

Electronic International Standard Serial Number (EISSN)

  • 1532-6551

Digital Object Identifier (DOI)

  • 10.1016/j.nuclcard.2007.06.115

Language

  • eng

Conference Location

  • United States