Radiographic observers' ability to recognize patient movement during cone beam CT

To assess radiographic observers' ability to recognize patient movement during cone beam CT and to decide early termination of the examination.

100 patients were video-recorded during cone beam CT examination. Patients' videos were cropped twice: fitting the active 20-s examination time or the initial non-radiation 3 s of the examination. x- and y-coordinates of pre-defined points marked on the patient's face were used to define the reference standard for movement in the 20-s videos. A sample of 40 non-moving and 20 moving patients was selected. Eight observers scored the videos. The 3-s videos were scored: 0, the patient did not move; 1, the patient moved and the examination should be terminated. The 20-s videos were scored: 0, the patient did not move; 1, the patient moved. Re-assessment of 15% of the videos provided intra-observer reproducibility. The 20-s videos were compared with the reference standard providing sensitivity and specificity values (movement/non-movement recognition). The scores of the 3-s videos were compared with the scores of the 20-s videos.

Intra- and interobserver reproducibility ranged from substantial to almost perfect for both videos. The 20-s videos allowed patient movement recognition with a high specificity and a medium to high sensitivity. The 3-s videos allowed early termination of the examination with a small number of incorrect positive scores. The majority of the patients scored as moving in the 20-s videos were detected in the 3-s videos.

By observing video recordings, trained observers are able to recognize patient movement during cone beam CT examination with high specificity and to decide an early termination of the examination.