Impact of a predefined mediastinal ROI on inter-observer variability of planar ¹²³I-MIBG heart-to-mediastinum ratio.

Aim: Purpose of this study was to assess the impact of mediastinal region of interest (ROI) definition on intra- and inter-observer variability in relation to collimator type.
Methods: Thirty-five subjects with CHF (80% men, mean age 66 ± 9 years, NYHA 2.4 ± 0.5, LVEF 29 ± 8.4%) were enrolled. 15 minutes and 4 hours post-injection (p.i.) of (123)I-MIBG, planar images were sequentially acquired with low energy high energy (LEHR) and medium energy (ME) collimators. In the first analysis, observer-defined mediastinal ROI was used. In the second analysis, a predefined mediastinal ROI was used. Intra- and inter-observer variability of late H/M was assessed using Lin's concordance coefficient (LCC).
Results: There was substantial agreement between all three observers using predefined mediastinum ROI. LCCs for LEHR were 0.98, 0.96, and 0.95, for ME 0.98, 0.97, and 0.97. However, observer-defined mediastinal ROI resulted in poor-moderate agreement. LCCs for LEHR were 0.82, 0.94, and 0.70, for ME 0.77, 0.91, and 0.80. Intra-observer analysis using predefined mediastinal ROI showed substantial agreement. LCC was 0.97 for LEHR and 0.96 for ME.
Conclusion: Predefined mediastinal ROI results in low intra- and inter-observer variability of late H/M and is, therefore, to be preferred over observer-defined mediastinal ROI. Intra- and inter-observer variability of late H/M is not influenced by collimator choice.