Modeling plan-related clinical complications using machine learning tools in a multiplan IMRT framework.
Modeling plan-related clinical complications using machine learning tools in a multiplan IMRT framework.
复制标题
DOI：
10.1016/j.ijrobp.2009.02.065
复制
发表时间：
2009-08-01
影响因子：
7
通讯作者：
Meyer, Robert R.
中科院分区：
文献类型：
--
作者：
Zhang, Hao H.;D'Souza, Warren D.;Si, Leyuan;Meyer, Robert R.
预测危险器官(OAR)并发症作为剂量-体积(DV)约束设置的函数，而不需要在多计划调强放疗框架中进行显式计划计算。通过改变OARS(多计划框架)上的剂量-体积约束(输入特征)来生成大量计划，并且将计划中的OARS实现的剂量-体积水平(计划属性)建模为施加的剂量-体积约束设置的函数。然后，通过单独使用施加的剂量-体积约束(特征)或结合建模的剂量-体积水平(计划属性)作为机器学习(ML)算法的输入，对每个计划的OAR并发症进行预测。这些ML方法被用来模拟头颈部和前列腺调强放疗后的两种OAR并发症，口干症和2级直肠出血。模型验证采用两次交叉验证，并报告平均误差。将所获得的剂量-体积值作为约束设置的函数进行建模的误差为0-6%。在头颈部病例中，唾液流率归一化为治疗前唾液流率的平均绝对预测误差为0.42%，95%可信区间为[0.41%，0.43%]。在前列腺病例中，预测2级直肠出血并发症的平均准确率为97.04%，95%可信区间为[96.67%，97.41%]。ML可用于在治疗计划期间预测OAR并发症，从而允许在计划框架内评估替代剂量-体积限制设置。
To predict organ-at-risk (OAR) complications as a function of dose-volume (DV) constraint settings without explicit plan computation in a multi-plan IMRT framework. A large number of plans were generated by varying the dose-volume constraints (input features) on the OARs (multi-plan framework), and the dose-volume levels achieved by the OARs in the plans (plan properties) were modeled as a function of the imposed dose-volume constraint settings. OAR complications were then predicted for each of the plans by using the imposed dose-volume constraints alone (features) or in combination with modeled dose-volume levels (plan properties) as input to machine learning (ML) algorithms. These ML approaches were used to model two OAR complications following head-and-neck and prostate IMRT, xerostomia and Grade 2 rectal bleeding. Two-fold cross-validation was used for model verification and mean errors are reported. Errors for modeling the achieved dose-volume values as a function of constraint settings were 0-6%. In the head and neck case, the mean absolute prediction error of the saliva flow rate normalized to the pre-treatment saliva flow rate was 0.42% with a 95% confidence interval of [0.41%, 0.43%]. In the prostate case, an average prediction accuracy of 97.04% with a 95% confidence interval of [96.67%, 97.41%] was achieved for Grade 2 rectal bleeding complications. ML can be used for predicting OAR complications during treatment planning allowing for alternative dose-volume constraint settings to be assessed within the planning framework.
登录
查看更多内容
影响因子：
3.8
作者：
Chen, Shifeng;Zhou, Sumin;Das, Shiva K.
通讯作者：
Das, Shiva K.
影响因子：
3.5
作者：
Craft, David;Bortfeld, Thomas
通讯作者：
Bortfeld, Thomas
DOI：
10.1016/j.ijrobp.2004.09.022
发表时间：
2005-02-01
影响因子：
7
作者：
Rosen, I;Liu, HH;Liao, ZX
通讯作者：
Liao, ZX
DOI：
10.1016/j.ijrobp.2007.08.019
发表时间：
2007-12-01
影响因子：
7
作者：
Craft, David;Halabi, Tarek;Bortfeld, Thomas
通讯作者：
Bortfeld, Thomas
影响因子：
3.5
作者：
Xing, L;Li, JG;Boyer, AL
通讯作者：
Boyer, AL