Inter-operator agreement and temporal reproducibility of automatically and manually generated multiplanar...

Automatic multiplanar reconstruction (Auto MPR) for head CT has mainly been evaluated for accuracy, though temporal reproducibility remains underassessed. We compared Auto MPR and manual MPR (Manual MPR) axial images, evaluating inter-operator agreement and temporal reproducibility separately, in 16 patients who underwent three non-contrast head CT scans within two weeks, each independently processed by seven technologists. After correcting in-plane translational offsets, spatial overlap of skull regions was quantified using the Dice similarity coefficient (DSC) and intersection over union (IoU). Auto MPR showed significantly higher inter-operator agreement (median DSC, 0.849 vs. 0.776; p < 0.001), whereas Manual MPR showed higher temporal reproducibility (median DSC, 0.850 vs. 0.792; p = 0.018), particularly in intracranial hemorrhage. Inter-operator agreement and temporal reproducibility represent independent quality dimensions of Auto MPR. We propose a diagnosis-dependent workflow: Auto MPR for single time-point, multi-technologist evaluations, and Manual MPR-or Auto MPR with manual verification-for longitudinal follow-up requiring precise comparison.