ABSTRACT
Objective:To investigate the performance of optical surface imaging (OSI) in the postmastectomy radiotherapy setup and to assess the effects of 3D printed silicone bolus on OSI detection precision.Methods:A retrospective analysis was conducted for 16 patients treated with left-sided postmastectomy radiotherapy (PMRT) in West China Hopital, Sichuan University from January to April, 2021. The setup errors of 16 patients without bolus detected using OSI (OSI no-bolus, OSI n) were obtained before error correction was conducted using cone-beam CT (CBCT). The correlation between OSI n and CBCT was analyzed, and then the diagnostic efficacy of OSI was assessed using the receiver operating characteristic (ROC) curves. The setup errors of six patients with 3D printed silicone bolus detected using OSI (OSI bolus, OSI b) were obtained through off-line image registration, and then the detection precision of OSI n and OSI b in the translational directions was compared. Results:The setup errors in the case of OSI n were highly correlated with CBCT in the translational direction ( r ≥ 0.80), but were weakly correlated in the rotation direction ( r < 0.40). In the ROC analysis, the area under the curve (AUC) in the y direction was the lowest and was in the order of AUC 5 mm ≥AUC 3 mm > 0.75 for any translational direction. The difference in the detection precision between OSI n and OSI b was not statistically significant in the x and z directions ( P > 0.05), but was statistically significant in the y direction ( Z = -2.56, P = 0.01). In the y direction, the systematic error of detection precision in the case of OSI b was 3.11 mm higher than that in the case of OSI n, and the random error of detection precision in the case of OSI b was 1.9 mm higher than that in the case of OSI n. Conclusions:OSI cannot yet substitute CBCT in the postmastectomy radiotherapy setup, but its detection error is still within the clinically acceptable range. The performance of OSI-assisted setup is expected to be further improved by mitigating the interference of factors such as bolus in the imaging path through operational training.