Abstract: Objective To evaluate the accuracy of 3D ultrasound calibration in image guided radiotherapy for prostate cancer by taking cone beam CT calibration as the gold standard， and to analyze the risk factors of accuracy. Methods From December 2018 to December 2021， 51 patients with prostate cancer from the Department of Radiation Oncology， First Affiliated Hospital of Hebei North University were selected as the study subjects. They received cone beam CT calibration based on bone and 3D ultrasound calibration based on soft tissue before fraction volumetric modulated arc therapy treatment three times a week. Taking cone beam CT calibration data as the gold standard， the Bland-Altman method was used to analyze the consistency of 3D ultrasound calibration data with the former. Taking 3 mm as the allowable threshold of accuracy， the calibration accuracy of 3D ultrasound relative to cone beam CT was evaluated. Logistic regression was used to analyze the risk factors affecting the accuracy of 3D ultrasound calibration. Results A total of 765 pairs of cone beam CT and 3D ultrasound calibration data were obtained from 51 patients in left-right， superior-inferior and anterior-posterior directions. The calibration data of 3D ultrasound and cone beam CT were （1.39±0.11） and （1.13±0.07） mm in the left-right direction， （1.98±0.20） and （1.61±0.12） mm in the superior-inferior direction， （2.68±0.48） and （1.78±0.27） mm in the anterior-posterior direction， respectively， with statistically significant differences （t=-6.42， P<0.001； t=-7.07， P<0.001； t=-7.34， P<0.001）. The analysis results of Bland-Altman showed that the consistency of calibration data of the two methods were acceptable in three directions. The number of pairs of 3D ultrasound relative to cone beam CT calibration data difference <3 mm in the three directions were 676 （88.37%） on the left-right direction， 604 （78.95%） on the superior-inferior direction， and 577 （75.42%） on the anterior-posterior direction. The factors with statistically significant differences in the left-right direction included age （χ²=18.27， P<0.001）， prostate volume （χ²=14.55， P<0.001）， Charlson comorbidity index （CCI）（χ²=8.01， P=0.005） and field range （χ²=11.30， P<0.001）. Age （OR=2.02， 95%CI： 1.90-3.39， P=0.010） and the field range （OR=1.45， 95%CI： 1.18-2.55， P=0.020） were the independent risk factors affecting the accuracy of 3D ultrasound calibration in the left-right direction. The factors with statistically significant differences in the superior-inferior direction included age （χ²=80.68， P<0.001）， body mass index （χ²=35.89， P<0.001） and field range （χ²=40.39， P<0.001）. Age （OR=1.49， 95%CI： 1.15-2.09， P=0.021） and the field range （OR=1.10， 95%CI： 1.01-1.90， P=0.034） were the independent risk factors affecting the accuracy of 3D ultrasound calibration in the superior-inferior direction. The factors with statistically significant differences in the anterior-posterior direction included age （χ²=46.07， P<0.001）， CCI （χ²=47.97， P<0.001） and field range（χ²=11.86， P=0.001）. Age （OR=1.91， 95%CI： 1.22-3.45， P=0.015） and the field range （OR=2.89， 95%CI： 1.45-3.90， P=0.001） were the independent risk factors affecting the accuracy of 3D ultrasound calibration in the anterior-posterior direction. Conclusion The consistency and accuracy of the calibration results of 3D ultrasound relative to cone beam CT are acceptable. It is necessary to consider the patient's age and field range to reduce the impact on accuracy before conducting 3D ultrasound calibration.

Key words: Prostatic neoplasms, Ultrasonography, Risk factors, Image-guided