Image registration accuracy of an in-house developed patient transport system for PET/CT+MR and SPECT+CT imaging

The aim of this study was to investigate the registration accuracy of a newly developed patient shuttle system that can integrate different scanners by patient transfer without repositioning for 'hardware'-based image fusion. We aimed to assess the registration accuracy of image fusion in two different settings: a trimodality PET/CT+MR system and a SPECT+CT system.

In this prospective study, 43 patients underwent either sequential PET/CT and MR (n=31) or sequential SPECT and diagnostic CT (D-CT) (n=12). A side-loading patient shuttle system was used for patient transport. For PET/CT+MR, hardware-only coregistration was performed and then validated with anatomical landmarks on CT and MR. SPECT+D-CT image fusion was performed with external cobalt-57 markers and manual fusion. Registration accuracy was analysed by anatomical landmarks on the attenuation correction CT and the D-CT.

For the PET/CT+MR system, the mean offset between original CT and MR images in all 31 patients was 8.1±5.7 mm in the X-axis, 5±4 mm in the Y-axis and 4.9±5.6 mm in the Z-axis. The validation of the cobalt-57 marker-assisted SPECT+D-CT fusion yielded offsets of 0.7±1.7 mm in the X-axis, 2.1±1.7 mm in the Y-axis and 0.8±1.8 mm in the Z-axis.

Sequential PET/CT+MR and SPECT+D-CT imaging using a dedicated patient shuttle system is feasible, resulting in mean offsets between data sets of 10.7 mm using the gantry laser system and 2.4 mm with fiducial markers.