Comparison of electronic portal imaging and cone beam computed tomography for position verification in children

To compare the accuracy of radiotherapy set-up using an electronic portal imaging device (EPID) versus megavoltage cone beam computed tomography (MV-CBCT) in paediatric patients.

In total, 204 pairs of EPID and MV-CBCT were carried out for 72 patients in the first 3 treatment days and weekly thereafter.

For the whole group, the mean systematic EPID set-up errors were 1.8 (±1.7), 1.6 (±1.3), 1.4 (±1.5) mm and 2.3 (±1.7), 1.6 (±1.3), 2.4 (±1.6) mm for MV-CBCT in the longitudinal, lateral and vertical directions, respectively, whereas the mean EPID random errors were 2.0 (±1.7), 1.4 (±1.5), 1.2 (±1.6) and 1.9 (±1.5), 1.5 (±1.3), 2.1 (±1.7) mm for MV-CBCT in the longitudinal, lateral and vertical directions, respectively. For systematic errors of head and neck patients, there was a statistically significant difference in the lateral and vertical directions (P=0.027, 0.003), whereas in the non-head and neck patients there was a statistically significant difference in the lateral direction only (P=0.031). In head and neck patients, the mean random errors were significantly different in the vertical and lateral directions, whereas in non-head and neck patients, they were significantly different in the vertical direction only. The larger values alternate between the two modalities. The systematic and random errors (detected by EPID and MV-CBCT) were significantly correlated in almost all direction in all tumour sites.

The comparison between set-up error in EPID and MV-CBCT was not in favour of any of the two modalities. However, the two modalities were strongly correlated but fairly agreed and the differences between the shifts reported were small and hardly influenced the recommended planning target volume margin.