文章快速检索 高级检索
 南方医科大学学报  2017, Vol. 37Issue (12): 1626-1631  DOI: 10.3969/j.issn.1673-4254.2017.12.11. 0

引用本文 [复制中英文]

CHEN Jiawei, CHEN Haibin, HE Qiang, LIAO Yuliang, ZHEN Xin. Rectal toxicity prediction based on accurate rectal surface dose summation for cervical cancer radiotherapy[J]. Journal of Southern Medical University, 2017, 37(12): 1626-1631. DOI: 10.3969/j.issn.1673-4254.2017.12.11.

文章历史

Rectal toxicity prediction based on accurate rectal surface dose summation for cervical cancer radiotherapy
CHEN Jiawei, CHEN Haibin, HE Qiang, LIAO Yuliang, ZHEN Xin
School of Biomedical Engineering, Southern Medical University, Guangzhou 510515, China
Supported by National Natural Science Foundation of China (81728016, 81571771)
Abstract: Objective To propose arectal toxicity prediction method based on deformable surface dose accumulation. Methods The clinical data were collected retrospectively from 42patients receiving radiotherapy for cervical cancer. With the first fraction as the reference, the other fractions of rectum surface were registered to the reference fraction to obtain the deformation vector fields (DVFs), which were used to deform and sum the fractional rectal doses to yield the cumulative rectal dose. The cumulative rectal dose was flattened via 3D-2D mapping to generate a 2D rectum surface dose map. Two dosimetric features, namely DVPs and DGPs were extracted. Logistic regression embedded with sequential forward feature selection was used as the prediction model. The predictive performance was evaluated in terms of the accuracy, sensitivity, specificity, and the area under the receiver operating characteristic (ROC) curve (AUC). Results Significant improvements for rectum surface DIR were achieved. The best predictive results were achieved by using both DVPs and DGPs as the features with a sensitivity of 79.5%, a specificity of 81.3% and an AUC of 0.88. Conclusion The proposed method is feasible for predicting clinical rectal toxicity in patients undergoing radiotherapy for cervical cancer.
Key words: cervical cancer    rectum toxicity    deformable registration    dose accumulation    logistic regression

1 资料和方法 1.1 研究对象

1.2 方法概述

1.2.1 TOP-DIR直肠表面点配准

TOP-DIR是基于TPS-RPM算法[21]的非刚性点配准方法：定义三维浮动点集$V = \left\{ {{{\vec v}_i} = \left( {v_i^x, v_i^y, v_i^z} \right)|i = 1, 2, \ldots, K} \right\}$和参考点集$X = \left\{ {{{\vec x}_j} = \left( {x_j^x, x_j^y, x_j^z} \right)|j = 1, \ldots, N} \right\}$，TPS-RPM算法通过最小化下面的能量函数E，迭代求解两点集之间的匹配矩阵M和映射函数f

 $\begin{array}{l} \arg \mathop {\min }\limits_{M, f} \left( {M, f} \right) = \sum\limits_{i = 1}^{K + 1} {\sum\limits_{j = 1}^{N + 1} {{m_{ij}}} } {\left\| {{{\vec x}_j}-f\left( {{{\vec v}_i}} \right)} \right\|^2} + \lambda {\left\| {Lf} \right\|^2} + \\ T\sum\limits_{i = 1}^K {\sum\limits_{j = 1}^N {{m_{ij}}} } \log {m_{ij}} + {T_0}\sum\limits_{j = 1}^N {{m_{N + 1, j}}} \log {m_{N + 1, j}} + \\ {T_0}\sum\limits_{i = 1}^K {{m_{i, K + 1}}} \log {m_{i, K + 1}} \end{array}$ (1)

 ${P_{ij}} = \frac{{{m_{ij}}{S_{ij}}}}{{\sum\limits_{k = 1}^N {{m_{ik}}} {S_{ik}}}}$ (2)
 ${S_{ij}} = \sum\limits_{s \in N_V^i} {\sum\limits_{t \in N_X^j} {{R_{ij}}} \left( {s, t} \right){m_{st}}}$ (3)

 $\alpha \left( {i, s;j, t} \right) = 1-\left| {\frac{{\left( {d\left( {i, s} \right)-d\left( {j, t} \right)} \right)}}{{\mathop {\max }\limits_{m \in N_V^i, n \in N_X^i} \left( {d\left( {i, m} \right)-d\left( {j, n} \right)} \right)}}} \right|$ (4)
 $\begin{array}{l} \beta \left( {i, s;j, t} \right) = \left( {1-d\left( {i, s} \right)/\mathop {\max }\limits_{m \in N_V^i} \left( {d\left( {i, m} \right)} \right)} \right)\\ \left( \begin{array}{l} 1-d\left( {j, t} \right)/\\ \mathop {\max }\limits_{n \in N_X^j} \left( {d\left( {j, n} \right)} \right) \end{array} \right) \end{array}$ (5)

1.2.2 剂量累加及3D-2D剂量映射

 图 1 3D-2D剂量映射示意图 Figure 1 Illustration of unfolding the 3D rectum surface dose(left) to a 2D rectum surface dose map (left).
1.2.3 剂量特征提取

1.2.4 直肠并发症预测模型

 $L\left( \theta \right) = \prod\limits_{i = 1}^m {P\left( {{y^i}|{x^i}} \right)} = \prod\limits_{i = 1}^m {\left( {{h_\theta }\left( {{x^i}} \right)} \right)}^{y^i} {\left( {1-{h_\theta }\left( {{x^i}} \right)} \right)^{1-{y^i}}}$ (6)
1.2.5 统计分析

2 结果 2.1 直肠表面点配准

 图 2 直肠形变较小（左）、较大（中）以及形变较复杂（右）3种情况下TOP-DIR的配准结果 Figure 2 Three examples of rectum TOP-DIR with small, large and complex deformation.

2.2 预测模型定量分析

 图 3 不同特征组合所建立的LR-SFS模型的ROC曲线对比分析 Figure 3 ROC analysis for different features and their combinations via LR-SFS.
3 讨论