Abstract
Experimental data, describing patient inter-fractional set-up errors in the clinical practice of breast radiotherapy, were exploited to simulate the performance in errors detection and correction of a constrained surface registration procedure, based on a hybrid configuration of control points (passive and laser). During 47 treatment sessions in three patients undergoing post-quadrantectomy radiotherapy, an opto-electronic localizer was used to acquire the three-dimensional coordinates of the hybrid control points, being two passive markers placed on selected skin landmarks on the sternum. Laser scanning technique was also applied for the acquisition of the 3-D surface model of the irradiated body area, which was used as reference for the automatic position correction procedure. A constrained surface registration algorithm was applied to estimate the rigid spatial transformation, describing the local errors affecting the control points. The improvement of the irradiation geometrical setup, by correcting the patient position according to the estimated spatial transformation parameters, was simulated. Results showed that the proposed surface registration method allowed us to detect and significantly (Wilcoxon signed rank analysis) reduce the initial misalignments, which exhibited overall median and 75th percentile values equal to 4.26 mm and 5.76 mm. Simulated residual errors dropped down to median and 75th percentile values measuring 2.95 mm and 3.87 mm, respectively. These results confirmed the high potentiality of surface registration techniques for the opto-electronic automatic patient positioning control in breast cancer radiotherapy.