计算机应用与软件2025，Vol.42Issue(7)：44-50,7.DOI:10.3969/j.issn.1000-386x.2025.07.007
基于影像组学的术前直肠癌淋巴血管侵犯状态预测
PREOPERATIVE LYMPHOVASCULAR INVASION STATUS PREDICTION IN RECTAL CANCER BASED ON RADIOMICS
摘要
Abstract
As for prediction of Lymphovascular invasion(LVI)before surgery,a classification model is constructed to predict the LVI status of rectal cancer patients based on the imaging omics method.212 patients with rectal cancer confirmed by pathological examination were studied,a variety of machine learning algorithms were used to screen the image features extracted by PyRadiomics,and the support vector machine algorithm was used to establish the image model.The logistic regression algorithm with Akaike information criterion as the evaluation index was used to screen the clinical features,and the logistic regression algorithm was used to establish the clinical model.Image features and clinical features were integrated to build the combined model using logistic regression algorithm.According to the results,in the training set and validation set,the model integrating imaging features and clinical features show the best predictive ability(training set AUC:0.954;validation set AUC:0.909),and the proposed model can predict LVI status of rectal cancer patients before surgery,which also can be used as an effective clinical tool to guide following personal treatment.关键词
直肠癌/淋巴血管侵犯/影像组学/逻辑斯特回归/支持向量机Key words
Rectal cancer/Lymphovascular invasion/Radiomics/Logistic regression/Support vector machine分类
信息技术与安全科学引用本文复制引用
吉祥,张茜,林予松,付芳芳,杨燕,王梅云..基于影像组学的术前直肠癌淋巴血管侵犯状态预测[J].计算机应用与软件,2025,42(7):44-50,7.基金项目
国家自然科学基金面上项目(81772009) （81772009）
河南省科技攻关计划项目(222102210281) （222102210281）
郑州市协同创新重大专项(20XTZX05015,20XTZX06013). （20XTZX05015,20XTZX06013）
