扇形束 CT 引导头颈部肿瘤放疗的智能摆位与人眼摆位精度比较.

Objective To compare the differences in set-up errors between human eye positioning and artificial intelligence (AI) positioning guided by fan-beam CT (FBCT) during radiotherapy for head and neck tumors (HNT) . Methods FBCT guided radiotherapy (FBCT-RT) was routinely used for HNT treatment on uRT-linac 506c accelerator. FBCT scans were carried out at specific intervals, including the first three fractions and then once a week. The set-up errors were analyzed in three dimensions. The second FBCT scan was performed using only AI positioning without visual calibration (experimental group), while the third FBCT scan was implemented using human positioning (control group). Self-compared t-test was used to compare the set-up errors and positioning time of the two groups. Analysis of variance was used to compare the set-up errors under AI positioning during the six-week course of radiotherapy. Thirty-five HNT patients were enrolled in the study from September 2022 to November 2022. Results The set-up errors in vertical direction (VRT),longitudinal direction (LNG) and lateral direction (LAT) direction were all statistically significant (P<0.05) between the two groups. It was found that set-up errors could be stably controlled with AI positioning during the first three weeks of radiotherapy for all patients. However, instability of VRT occurred during the last three weeks of radiotherapy for malignancies (P<0.05), althrough set-up accuracy remained stable for orbital tumors. Conclusions AI positioning guided by FBCT not only saved set-up time but also reduced set-up errors during earlier stages of radiotherapy. Compared with other HNTs, orbital benign tumors benefited most from AI positioning due to lower set-up errors. [ABSTRACT FROM AUTHOR]