Quantitative analysis of striatal 123I-Ioflupane uptake with HERMES Affinity software

Introduction: DaT imaging with ¹²³I-Ioflupane is a common diagnostic tool for Parkinson’s Disease (PD). However, the visual interpretation of these images can be challenging, leading to variations in reporting. Currently available HERMES Affinity software makes quantifying ¹²³I-Ioflupane uptake feasible. The objectives of this study were (1) to determine the feasibility and inter-operator variability of quantitative analysis of striatal uptake with the new software and (2) to compare the visual assessments with the quantitative data.

Methods: Retrospective review on consecutive patients with suspected or diagnosed PD referred to ¹²³I-Ioflupane SPECT in an academic centre between 2021-2024 was conducted. Images were acquired at 3-4 hours after injection of 155.3±4.8 MBq of ¹²³I-Ioflupane. Striatal uptakes were assessed both visually and quantitatively. For quantification, the background count density (counts/voxel) was measured at the occipital lobes at the level of inferior striatal uptake on the axial images using 50 ml spheres. The threshold for the region of striatal uptake was set at twice background count density. The striatal volume for each side was segmented with Affinity and the total counts were measured. The net activity was determined by total counts*(1-background mean counts/striatal mean counts)/dose. Both background and striatal measurements were performed by two readers. The visual rating scores were obtained from the study reports and assessed by two readers (a physician and a technologist) blinded to clinical information, with the normal uptake defined as the highest score (score=16) and no uptake as 0. Inter-reader variability of quantitative measurements was assessed using Pearson’s correlation and Bland-Altman plot. The correlation between the quantitative data and the visual score, as well as visual score inter-reader variability was assessed using Spearman’s correlation.

Results: 16 patients (44% male, ages 56-72) were studied, from which 9 had 2 and 7 had 3 scans taken at different time intervals, totalling to 39 scans. Using the Affinity viewer the background and the striatal mean densities were measured to be 85±14, and 222±40, respectively. Total counts in the right and left striata were 44618±23392 and 40534±15723, respectively, corresponding to net activities of 184±102 counts/MBq in the right and 166±68 counts/MBq in the left striatum. Excellent inter-reader agreement was observed for the net activity in both the right (ρ=0.996, p<0.001) and left (ρ=0.991, p<0.001) striata. Bland-Altman plot indicated no bias with good agreement. The total visual score (combined the right and left striata) was 8.0±2.9 and 7.7±2.4 from physician and technologist readers and 10.0±2.1 from the report. The correlation between the total striatal net activity and the visual score was strong for physician reader (ρ=0.741, p<0.001), moderate for technologist reader (ρ=0.569, p<0.001) and weak from the report (ρ=0.398, p=0.013). For the correlation of visual scores, there was weak to moderate correlation between physician and technologist (ρ=0.615, p<0.001), physician and report (ρ=0.452, p=0.004), and technologist and report (ρ=0.558, p<0.001).

Conclusions: Quantifying ¹²³I-Ioflupane uptake is feasible and applicable to daily practice. There was excellent inter-reader correlation in striatal uptake quantification of 123I-Ioflupane, suggestive of robust nature of using Affinity. The quantitative striatal activity was strongly correlated with the physician reader’s visual score but weakly with the report, which could partially be due to inter-reader variability among reporting physicians. In addition, improving diagnostic accuracy could be achieved by paying attention to the background activity, as the striatal activity could be overestimated without adjusting (decreasing) background uptake on visual assessment. We anticipate that quantification data from the current study will aid reporting accuracy in practice.