Comparison of quantitative Krenning Scores with visual assessment in 99m Tc-EDDA/HYNIC-TOC SPECT-CT.

Abstract

Purpose: The aim of this study is to assess inter-observer variability of the Krenning Score for 99m Tc-EDDA/HYNIC-TOC single photon emission computed tomography (SPECT)-computed tomography (CT) images and compare against quantitative metrics obtained from tumour and physiological uptake measurements.

Methods: Thirty-two patients with 117 lesions visible on 99m Tc-EDDA/HYNIC-TOC SPECT-CT were scored by two expert observers using the Krenning Score. Five observers with less extensive experience also scored the lesions on visual assessment. Inter-observer agreement and comparison to the expert consensus was tested. Three observers made quantitative measurements of the lesions and physiological uptake, with intra-observer and inter-observer variation investigated. Assessment of agreement between quantitative metrics and the expert visual consensus was also made.

Results: Inter-observer agreement for visual assessment was 44.3% for proportions of agreement and 0.576 for Fleiss' Kappa, whilst for the best-performing quantitative metric the inter-observer Fleiss' Kappa was equal to 1. The agreement with expert consensus for the best-performing visual observer was 89.8% for percentage of agreement and 0.914 for Cohen's Kappa, similar to the best-performing quantitative metric (a derived quantitative Krenning Score) at 86.4% and κ  = 0.877. Standardised uptake value maximum (SUV max ) also showed similar levels of agreement at 85.1% and κ  = 0.871.

Conclusion: A derived quantitative Krenning Score, or alternatively SUV max , can provide similar levels of agreement with an expert consensus Krenning Score as visual assessment, with reduced inter-observer variability. Quantification can deliver greater consistency in scoring of 99m Tc-EDDA/HYNIC-TOC images over visual assessment, an important factor when imaging is used to determine patient eligibility for peptide receptor radiotherapy.