基于近红外散射光谱的宫颈癌放疗疗效评估

Evaluation of the Efficacy of Radiotherapy for Cervical Cancer Based on Near-Infrared Scattering Spectroscopy

背景与目的:利用近红外散射光谱的特征分析技术,在放疗过程实时观察到病人病灶治疗情况,实现便捷、无损的辅助疗效评估。材料与方法:通过由卤素光源、光纤光谱仪、双光纤探头和上位机软件组成的数据采集系统,采集宫颈癌治疗前、放疗期间和放疗后的散射光谱,将这些光谱数据进行一系列数据处理、分析,再随机分为训练集(70%)、验证集(15%)和测试集(15%),并利用MATLAB自带的神经网络工具箱进行建模,通过测试集来验证该模型的有效性。临床试验经筛选后得到的样本有74例,其中有14例治疗前,40例治疗中,20例治疗后。结果:BP神经网络算法结合光谱数据的方法可以用于识别宫颈癌放疗前、放疗过程中与放疗治愈后的宫颈在体组织,能够初步进行宫颈癌辅助疗效评估且识别准确率在90%左右。结论:本研究表明通过对宫颈癌组织治疗过程中光谱的数据分析,可以实现治疗阶段的基本判定,为宫颈癌的放疗疗效评估寻找到一种更加便捷、简单的辅助方法。

Background and objective:Using the feature analysis technology of near-infrared scattering spectroscopy,it is possible to observe the treatment status of the patient’s focus in real time during the radiotherapy process,which can realize convenient and non-destructive auxiliary efficacy evaluation.Material and methods:A data acquisition system consisting of a halogen light source,optical fiber spectrometer,dual optical fiber probes and PC software is used to collect the scattering spectra of cervical tissue before cancer treatment,spectral data during radiotherapy and spectral data after radiotherapy,and perform a series of data processing and analysis on these spectral data.Then these spectral data are randomly divided into training set(70%),validation set(15%)and test set(15%),and use MATLAB’s own neural network toolbox for modeling,and use the test set to verify the effectiveness of the model.There are 74 samples obtained after screening in clinical trials,of which 14 were before treatment,40 were during treatment,and 20 were after treatment.Results:The method of BP neural network combined with spectral analysis is feasible to identify in vivo tissues before,during and after radiotherapy for cervical cancer.It can be used to evaluate the auxiliary efficacy of cervical cancer and the recognition accuracy is about 90%.Conclusion:This study shows that through the analysis of the spectral data during the treatment of cervical tissue,the basic judgment of the treatment stage can be realized,and a more convenient and simple auxiliary method for the evaluation of the efficacy of cervical cancer can be found.