人工智能技术用于肺癌放射治疗靶区勾画中的临床效果研究

Study on clinical effect of artificial intelligence technique in delineating target volume of radiotherapy for lung cancer

目的:探讨人工智能(AI)技术在肺癌患者放射治疗中靶区勾画的临床效果。方法:选择2021年9月至2023年3月在萍乡市人民医院行放疗治疗的60例肺癌患者,采用随机信封法将其分为对照组和观察组,每组30例。对照组按照传统方法勾画靶区,观察组采用深度学习技术进行训练,输出和利用UNet网络模型完成患者放疗靶区的自动勾画,比较两组近期疗效、计划靶区体积(PTV)、靶区辐射剂量、危及器官(OAR)体积和剂量、生存期及不良反应发生率。结果:观察组干预后客观缓解率(ORR)为70.0%(21/30),高于对照组的46.67%(14/30),其差异有统计学意义(χ2=5.691,P<0.05);观察组内靶区(ITV)及计划靶区辐射剂量低于对照组,差异有统计学意义(t=4.591、4.934,P<0.05);观察组正常肺组织接受放射剂量20 Gy、5Gy的体积百分比(V20、V5)、双肺接受的平均肺部剂量(MLD)及脊髓1cc体积所受照剂量(D1cc)辐射剂量均低于对照组,差异有统计学意义(t=5.249、4.571、6.092、5.339,P<0.05);两组放疗过程中不良反应发生率比较,差异无统计意义(P>0.05)。结论:AI技术用于肺癌放疗中靶区勾画,能够提高患者ORR,有助于降低PTV、D95及适形指数,减少OAR体积和剂量,且未增加不良反应发生率。

Objective:To investigate the clinical effect of artificial intelligence(AI)technique in delineating target volume for patients with lung cancer during radiotherapy.Methods:A total of 60 patients with lung cancer who received radiotherapy in Pingxiang People's Hospital from September 2021 to March 2023 were selected,and they were divided into control group and observation group by random envelope method,with 30 cases in each group.The control group outlined target volume as conventional method.The observation group adopted deep learning technique to conduct train,and then,UNet network model was output and was used to complete automatic delineation for the target volume of radiotherapy for patients.The near-term efficacy,planning target region volume,radiation dose of target volume,volume and dose of organ at risk(OAR),survival time and incidence of adverse reactions were compared between two groups.Results:The objective relief rate(ORR)of observation group was 70.0%(21/30)after intervention,which was higher than that[46.67%(14/30)]of control group,and the difference was statistically significant(x2=5.691,P<0.05).The radiation doses of internal target volume(ITV)and planning target volume in observation group were lower respectively than those in control group(t=4.591,4.934,P<0.05),and the differences of them were significant,respectively.The volume percentages(V20,V5)of the exposed radiation dose that were higher than 20 Gy and 5 Gy in normal lung tissue,the exposed mean lung dose(MLD)of bilateral lungs and the exposed dose of 1cc volume(D1cc)of bilateral lungs in observation group were all lower than those in control group,the differences were statistically significant(t=5.249,4.571,6.092,5.339,P<0.05),respectively.There was no statistical significance in the incidence of adverse reaction between two groups(P>0.05).Conclusion:The application of AI technique in delineating target volume of radiotherapy for lung cancer can improve ORR,which is helpful to decrease the planning target volume,D95 and conformal index,and reduce the volume and dose of OAR.It does not increase the incidence of adverse reactions.