中国中西医结合影像学杂志2025，Vol.23Issue(3)：285-290,296,7.DOI:10.3969/j.issn.1672-0512.2025.03.004
临床指标联合多期相CT影像组学列线图模型对急性胰腺炎1年内复发的预测价值
Predictive value of a nomogram model based on clinical indicators combined with multi-phase CT radiomics for acute pancreatitis recurrence within one year
摘要
Abstract
Objective:To investigate the clinical value of a nomogram model integrating clinical indicators and multi-phase CT radiomics in predicting the recurrence of acute pancreatitis(AP)within one year.Methods:The clinical and imaging data of 155 patients with first-episode AP were retrospectively analyzed.After clinical follow-up for one year,155 patients were divided into two groups,a recurrent acute pancreatitis(RAP)group(39 cases)and a non-RAP group(116 cases).With a ratio of 7∶3,108 cases were assigned to the training cohort(27 cases with RAP,81 cases with non-RAP),47 cases were assigned to the validation cohort(12 cases with RAP,35 cases with non-RAP).The entire pancreas was delineated using a semi-automatic segmentation method.Feature selection was performed using Pearson correlation analysis,univariate analysis and LASSO algorithm.The support vector mechanism(SVM)was applied to establish radiomics models(plain scan model,arterial phase model,venous phase model)for predicting AP recurrence within one year.Logistic regression analysis was used to identify independent risk factors from clinical indicators for constructing clinical models.These clinical indicators were combined with radiomics scores of the combined radiomics model to develop a nomogram model.The predictive efficiency of the model was evaluated using ROC curve analysis,while decision curve analysis assessed its clinical application value.Results:Male and hyperlipidemia were identified as the independent risk factors,and the clinical model constructed by the 2 factors for predicting recurrence demonstrated an AUC of 0.764(sensitivity of 62.96%,specificity of 77.78%)in the training cohort and 0.670(sensitivity of 58.33%,specificity of 80.00%)in the validation cohort.The radiomics combined model based on multi-phase CT demonstrated AUCs of 0.883 and 0.850 in the training cohort and the validation cohort,respectively,which were higher compared to the plain scan model(AUC of 0.827 and 0.755),the arterial phase model(AUC of 0.785 and 0.771)and the venous phase model(AUC of 0.746 and 0.745).Furthermore,there were statistically significant differences between the combined radiomics model and both the arterial phase model and the venous phase model in the training cohort.The AUCs of the nomogram model constructed based on hyperlipidemia and radiomics scores in predicting AP recurrence within one year were 0.888 and 0.869 in the training cohort and the validation cohort,respectively,both of which were higher than those of the clinical model and the combined radiomics model.Conclusion:The nomogram model,incorporating multi-phase CT radiomics and clinical factors,demonstrates significant clinical utility in predicting AP recurrence within one year.关键词
急性胰腺炎/影像组学/体层摄影术,X线计算机/复发Key words
Acute pancreatitis/Radiomics/Tomography,X-ray computed/Recurrence引用本文复制引用
刘杰,谢玉海,鲍传飞,哈传传,李晓迪,郭莉,张子阳,赵梓涵,朱启航,崔涛..临床指标联合多期相CT影像组学列线图模型对急性胰腺炎1年内复发的预测价值[J].中国中西医结合影像学杂志,2025,23(3):285-290,296,7.基金项目
北京医学奖励基金会睿影基金资助项目(YXJL-2023-0227-0111) （YXJL-2023-0227-0111）
皖南医学院校级科研项目(WK2024JXYY086). （WK2024JXYY086）
