Study on the relationship between CT image texture changes of parotids and acute xerostomia during radiation treatment for head and neck cancer
10.3760/cma.j.issn.0254-5098.2019.04.004
- VernacularTitle:基于放疗过程中腮腺图像纹理特征的变化早期预测头颈部肿瘤放射性口干症
- Author:
Zhumin YAN
1
;
Jingqiao ZHANG
;
Xiaojian CHEN
;
Long HAI
;
Xueming SUN
;
Hui WU
Author Information
1. 郑州大学附属肿瘤医院放疗科 450008
- Keywords:
Parotid gland;
Radiation-induced xerostomia;
Texture;
CT image;
Early prediction
- From:
Chinese Journal of Radiological Medicine and Protection
2019;39(4):262-267
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate the relationship between parotid image texture and acute radiation xerostomia (grade) during radiotherapy in patients with head and neck cancer.The mathematical model was established to predict the severity of radiation dry mouth in the early stage.Methods 23 patients with head and neck cancer treated with radiotherapy were observed.The degree of xerostomia was evaluated according to RTOG criteria.The weekly validated CT images of these patients during radiotherapy were collected and transmitted to the MIM system to outline the parotid gland,and an internal analysis program was developed in MATLAB (R2013a).The changes of texture features of weekly parotid CT images during radiotherapy were analyzed,including mean CT value (MCTN),standard deviation (STD),skewness,kurtosis,entropy and volume.The mathematical model was established,and the KNN method was used to optimize the model and predict the level of xerostomia.Results There was no significant correlation among the changes of MCTN,volume and the degree of xerostomia (P ＞ 0.05).However,according to the weekly changes of MCTN and volume,the model was established to predict the grade of xerostomia with an accuracy of 99％.Conclusions The changes of parotid gland MCTN and volume were significantly correlated with acute radiation xerostomia during radiotherapy for head and neck cancer,and the MCTN changes can be used to predict the severity of xerostomia in the early stage.
