Abstract
2088
Objectives The aim of this study was to develop a semi-automatic method for standardizing the size and positioning of myocardial and mediastinal regions of interest (ROI) in cardiac I-123 MIBG imaging and to improve the reproducibility of heart-to-mediastinum (H/M) ratio.
Methods Thirty-seven patients (mean age 53 ± 20 years, 19 male) underwent both early and delayed I-123 MIBG scans. The H/M ratio was calculated by both semi-automatic and manual methods and assessed for intra- and inter-observer variability. The software algorithm automatically determined a rectangular mediastinal ROI using information from I-123 MIBG uptake of the heart, lung, liver and thyroid when a circular cardiac ROI was manually set. In the manual method, polygonal and rectangular ROIs were set on the heart and upper mediastinum, respectively.
Results In the intra-observer study, the intra-class correlation coefficient (ICC) from the semi-automatic method showed excellent reproducibility for early and delayed imaging (ICC = 0.99, 0.99, respectively). The Bland-Altman plots demonstrated better agreement using the semi-automatic method than that in the manual method (mean difference (MD): 0.00 ± 0.05 vs. -0.03 ± 0.16). The inter-observer correlation and agreement were also good using the semi-automatic method (ICC=0.92 and 0.97 for early and delayed, respectively, MD = -0.04 ± 0.13).
Conclusions The H/M ratio derived from the semi-automatic method yielded high reproducibility in the intra- and inter-observer MIBG studies