Ejection fraction (EF) in gated blood pool studies: effect of collimation and processing

Full text: Gated Blood Pool studies are performed to monitor left ventricular ejection fraction at baseline, during and post chemotherapy. Variations in EF results were queried on several occasions especially for serial scans. The study was set up to assess reproducibility and inter-operator variations for different collimation and processing programme used. A cohort of thirteen patients was selected over a period of two months and scanned twice on the same day using 1. Low Energy High Sensitivity and 2. Low Energy High Resolution collimators. Each scan was then processed using 1. Siemens Manual 2. Siemens Semi-automatic, and 3. St Louis Manual EF evaluation programs. The data was processed by six nuclear medicine technologists blinded to the previous results. Each technologist analysed each patient using different processing a few days apart. The two factor analysis of variance with replication (ANOVA) showed that the effect of using different collimation on EF was insignificant (P>0 4). There was a significant difference in EF value between processing programmes used (P < 0002). There was no interaction between collimation and processing program used (P>0 4). As expected, manual modes of processing showed increased inter-operator variance for each type of collimation. However in the data (25 scans and 450 EF calculations) semi-automatic processing results showed that only 4 5% (18-30% for manual) were different from mean by 5-10%. No semi automatic result was different from mean by 10% or more, while for manual processing frequency was 25%. Copyright (2002) The Australian and New Zealand Society of Nuclear Medicine Inc