基于多参数磁共振成像影像组学在直肠癌新辅助治疗效果预测的价值研究

打开文本图片集

Value multiparametric MRI radiomics in predicting neoadjuvant therapy response in rectal cancer Ke Nan， Liu Kai， Yang Xian， Chen Jiao， Zhang Haiying， Fan Bin， Zuo Peng. ， ，

【Abstract】ObjectiveTo explore the value multiparametric MRI radiomics in predicting neoadjuvant therapy（NAT）response inrectal cancer.MethodsAretrospectiveanalysiswas conductedon8Opatientswith rectal cancerwho underwent NATand surgical treatment inour hospital from January 2O18to December 2020. Based on the pathological stage after NAT，the patients were divided into a good response group （ypTo-2NM）and a poor response group （ypT3-4 or N+ or M+ ）.Radiomics features multiparametricMRI （including T2WI，DWI and DCE-MRI） before NAT were extracted，including morphology，gray level statistics，texture，wavelet，and radiogenomics.The random forest algorithm was used to screen out radiomics features related to NAT response and builda prediction model.Receiver operating characteristic（ROC） curve analysiswasused to evaluate the predictive performance the model.ResultsThere were 32 cases in the good response group and48 cases in thepoorresponse group.Therandom forest algorithmselected15 radiomic features andconstructed aprediction modelbasedonmultiparametricMRI radiomics forNAT response.The accuracy the model in the training set and the test set was 94% and 90% ，respectively，the sensitivity was 94% and 88 % ，respectively，the specificity was 94% and 92% ，respectively，and the area under the curve （AUC） was O.969 and O.925，respectively. ConclusionMultiparametric MRI radiomics can provide predictive information for NAT response in rectal cancer， which helps in individualized treatment planning.

【Key words】Rectal neoplasms;Neoadjuvant therapy;Multiparametric magnetic resonance imaging; Radiomics；Prediction model

直肠癌是消化道常见恶性肿瘤，新辅助治疗（NAT）是综合治疗的重要组成部分，但患者对NAT的反应存在个体差异。（剩余6012字）