Performance of 18F-Florbetaben PET Image Reading Training in Japanese language.

Objectives The efficacy of the visual assessment of 18F-Florbetaben (FBB) positron emission tomography (PET) scans against histopathology confirmation of presence or absence of β-amyloid neuritic plaque in the brain has been previously established [1]. The aim of this study was to assess the performance of the 18F-FBB PET image reading training in the Japanese language, as it is intended to be used as part of the future training for nuclear medicine physicians in Japan.

Methods Five blinded Japanese readers without previous experience reading FBB images were trained in-person by an expert Japanese nuclear medicine physician using the 18F-FBB PET image reading training materials in the Japanese language. PET images of 50 subjects (mostly Japanese) from a number of studies were pooled (n=17 patients diagnosed with Alzheimer’s disease, n=17 Healthy Volunteers and n=16 subjects with Mild Cognitive Impairment), and randomly assigned for blinded visual assessment using the image assessment criteria defined by the sponsor company. These images were previously assessed with the same criteria by 5 independent non-Japanese, electronically trained blinded readers as 25 positive and 25 negative scans, which was used as reference. The final assessment was binary as positive (β-amyloid present) or negative (β-amyloid absent). Percent agreement between the results of each Japanese reader with the majority results of the non-Japanese readers was separately calculated. Percent agreement among majority reads from both groups was calculated. Inter-reader agreement was assessed by calculating Fleiss’s kappa across all five readers and Cohen’s kappa between each pair of readers. Intra-reader agreement was assessed by calculating Cohen’s kappa values per reader for a re-read of 10 randomly selected cases from the image data set.

Results The percent agreement between the majority read of the non-Japanese readers and the majority read of the Japanese readers was 100.0% (95% CI: 92.9 - 100.0). A very high agreement between individual Japanese readers and the majority read of non-Japanese readers was observed (mean (range): 96.4 (86-100) %); (100% (n=3), 96% (n=1) and 86% (n=1)). Excellent inter-reader agreement was observed; kappa across 5 blinded readers was 0.87 (95% CI: 0.78 - 0.96). The average inter-reader agreement between reader pairs was 0.87 (range: 0.72 - 1.00). The mean percent of intra-reader agreement was 90% (range: 70-100%). Mean intra-reader variability for each of the five blinded readers was 0.78 (range: 0.34-1.00).

Conclusions This study confirms that readers taught by the in-person training in the Japanese language perform similar to non-Japanese readers in the visual assessment of FBB scans. The visual assessment method and the in-person training in the Japanese language is appropriate for clinical use in Japan. 1. Sabri, O., et al., Florbetaben PET imaging to detect amyloid beta plaques in Alzheimer disease: Phase 3 study. Alzheimer's & Dementia, 2015. 11(8): p. 964-974.