The most striking observation was that human specific to non-specific striatal [
123I]FP-CIT binding ratios on the NeuroFocus system were found to be slightly higher (~ 20%) compared to the InSPira. This was not observed in the phantom study. In general, differences up to 30% between cameras are not uncommon [
3] and can be even greater when processing steps differ. It is unlikely that these differences were (partly) caused by binding/wash-out differences 3- and 4-h post-injection, since an opposite effect would be expected; slightly
higher specific to non-specific striatal [
123I]FP-CIT binding ratios in healthy controls were found at 4 h compared to 3 h post-injection in a multi-center study [
3]. It is therefore probable that these differences are caused by system performance. Unfortunately, the replacement of the NeuroFocus was scheduled with the arrival of the InSPira and was dismantled. For the acquisition of the phantom and patient scans presented in this paper, there was limited time where the two systems were both available. Therefore, we were not able to examine extensively and directly the difference in system performance and acquisition/processing differences between both systems. In a previous paper from our group, however, the technical performance of the InSPira system was described [
2]. This paper reported on scanner performance characteristics, such as spatial resolution and recovery. Since recovery was very similar between the two systems, it is not likely that this caused the observed differences. It should be noted that no correction for resolution recovery was applied for either systems. Another difference between the systems is that the reconstruction algorithm of the NeuroFocus corrects for background count rate by default. Unfortunately we do not have detailed insights into how the reconstruction algorithm incorporates this background rate and can therefore only hypothesize about possible effects. Subtraction of a certain constant value is expected to artificially enhance striatal ratios, particularly high ratios. Indeed, it was observed that increasing background count rate in the reconstruction for NeuroFocus phantom data was in line with this hypothesis, but only accounted for a few percent difference. Therefore, we believe that this might only partly explain the higher ratios observed on the NeuroFocus compared to the InSPira. Perhaps most important, the collimation of the two systems could have an impact on the image quality; scatter and energy penetration could affect quantitation. Interestingly, for clinical data strongest differences were observed. These effects were not so apparent for the phantom data. This seems to indicate that activity outside the field of view, resulting from activity distribution in the body of the participants, may explain differences between the systems. Indeed, a comparatively higher background was observed for the InSPira patient data, when compared to the NeuroFocus patient data. It may therefore be hypothesized that the InSPira may be more affected by septal penetration than the NeuroFocus, which may explain the lower striatal binding ratios observed for the Inspira system in the patient studies. It should be noted that no correction for scatter was performed for either of the systems. Finally, acquisition and processing differences between the systems, such as smoothing parameters, slice thickness, and resolution, could have influenced quantification by affecting automatic registration in BRASS. However, we feel that this does not likely explain the quantitative differences, since visual inspection of registration of phantom and patient data to the template using BRASS seemed successful for scans of both systems.
We are confident that although these differences between the systems exist, this will not affect diagnostic accuracy when visual assessment of DAT SPECT images is done together with quantitative assessment using an appropriate reference database for that specific system.