基于高分辨T2WI影像组学判定直肠癌术前T分期

Preoperative T staging of rectal cancer based on radiomics of high resolution T2WI

目的探讨基于高分辨T2WI提取的影像组学特征判定直肠癌术前T分期的应用价值。方法回顾性分析191例术后病理证实为直肠腺癌患者的临床及影像学资料,在术前高分辨T2WI上勾画整个肿瘤为ROI,通过一般线性LASSO分析提取14个影像组学特征,并构建影像组学模型。根据病理结果,将T0~T2期归为早期,T3~T4期归为局部进展期,将患者分为训练组(n=134)和验证组(n=57),对模型进行训练和验证。绘制ROC曲线及决策曲线(DCA),评估模型的可预测性。结果191例直肠癌中,早期68例,局部进展期123例。在训练组和验证组中,影像组学预测模型区分早期与局部进展期直肠癌的AUC为0.927、0.885,准确率为0.851、0.842,特异度为0.917、0.700,敏感度为0.814、0.919,F1值为0.861、0.829,Brier值为0.149、0.159。结论术前以基于高分辨T2WI构建的影像组学模型判定直肠癌T分期具有可行性。

Objective To explore the application value of radiomics features extracted from high-resolution T2WI for preoperative T staging of rectal cancer. Methods Clinical and imaging data of 191 patients with pathologically confirmed rectal cancer were retrospectively analyzed. Whole tumor ROI was delineated on high-resolution T2WI, and 14 imaging features were extracted using LASSO analysis. The patients were divided into early stage (T0-T2) and local advanced stage (T3-T4) according to the pathological results, and then were divided into training group ( n =134) and testing group ( n =57) to train and test the model. ROC curve and decision curve analysis (DCA) were used to assess the predictability of the model. Results Tumors in 68 cases were in early stage, while in 123 cases were in the local advanced stage. In training and testing groups, AUC of the predictive model in distinguishing early and local advanced stage of rectal cancer was 0.927 and 0.885, accuracy was 0.851 and 0.842, specificity was 0.917 and 0.700, sensitivity was 0.814 and 0.919, F1 was 0.861 and 0.829, Brier was 0.149 and 0.159, respectively. Conclusion Using radiomics model constructed from high-resolution T2WI to preoperatively evaluate T stage of rectal cancer is feasible.