Set-up errors in head and neck cancer patients treated with intensity modulated radiation therapy: Quantitative comparison between three-dimensional cone-beam CT and two-dimensional kilovoltage images

To compare the patient set-up error detection capabilities of three-dimensional cone beam computed tomography (3D-CBCT) and two-dimensional orthogonal kilovoltage (2D-kV) techniques.

3D-CBCT and 2D-kV projections were acquired on 29 head-and-neck (H&N) patients undergoing Intensity Modulated Radiotherapy (IMRT) on the first day of treatment (time 0) and after the delivery of 40 Gy and 50 Gy. Set-up correction vectors were analyzed after fully automatic image registration as well as after revision by radiation oncologists. The dosimetric effects of the different sensitivities of the two image guidance techniques were assessed.

A statistically significant correlation among detected set-up deviations by the two techniques was found along anatomical axes (0.60 < ρ < 0.72, p < 0.0001); no correlation was found for table rotation (p = 0.41). No evidence of statistically significant differences between the indications provided along the course of the treatment was found; this was also the case when full automatic versus manually refined correction vectors were compared. The dosimetric effects analysis revealed slight statistically significant differences in the median values of the maximum relative dose to mandible, spinal cord and its 5 mm Planning Organ at Risk Volume (0.95%, 0.6% and 2.45%, respectively), with higher values (p < 0.01) observed when 2D-kV corrections were applied.

A similar sensitivity to linear set-up errors was observed for 2D-kV and 3D-CBCT image guidance techniques in our H&N patient cohort. Higher rotational deviations around the table vertical axis were detected by the 3D-CBCT with respect to the 2D-kV method, leading to a consistent better sparing of organs at risk.