基于CCTA的冠状动脉周围脂肪组织影像组学诊断不稳定型心绞痛

Diagnosis of unstable angina pectoris based on CCTA radiomic of pericoronary adipose tissue

目的:探究基于冠状动脉CT血管成像(CCTA)的冠状动脉周围脂肪组织(PCAT)影像组学结合临床及影像特征对稳定型心绞痛(SA)及不稳定型心绞痛(UA)的诊断价值。方法:回顾性分析187例心绞痛患者的临床影像资料,包括SA患者(92例)及UA患者(95例),按7∶3比例分为训练组(130例)及验证组(57例)。通过提取并筛选右冠状动脉(RCA)及左前降支(LAD)近端PCAT影像组学特征,且量化RCA近端脂肪衰减指数(FAI),建立FAI模型及3种影像组学模型[RCA、LAD、RCA联合LAD(联合模型)]。运用多因素logistic回归筛选临床影像资料并结合影像组学评分(Rad-score)建立综合模型,然后采用逻辑回归(LR)及随机森林(RF)两种分类器分别构建上述5种模型。以受试者工作特征(ROC)曲线评价模型的诊断效能,并结合校准曲线评估校准度。结果:RCA联合LAD近端PCAT共筛选9个最优影像组学特征用于计算Rad-score,糖化血红蛋白(HbA1b)、冠状动脉树中是否存在高危斑块(HRP)及Rad-score为诊断SA及UA的独立危险因素。与LR模型相比,RF模型鉴别性能更高且于验证组中具有统计学意义(P<0.05)。在RF模型中,3种影像组学模型诊断效能无差异,但是综合模型训练组AUC 0.89,高于联合模型训练组AUC 0.86,P=0.004。校正曲线显示3种影像组学模型及综合模型的校准度良好。结论:基于CCTA的PCAT影像组学模型诊断UA优于FAI模型,将PCAT放射组学与临床及影像特征(HbA1b、HRP)相结合,可进一步提高UA的诊断能力。

Objective To investigate the diagnostic value of radiomics analysis of pericoronary adipose tissue(PCAT)captured by coronary computed tomography angiography(CCTA)combined with clinical and imaging signature for stable angina(SA)and unstable angina(UA).Methods The clinical and CCTA imaging data of 187 patients with angina pectoris,comprising of 92 SA patients and 95 UA patients,were retrospectively analyzed.The patients were randomly divided into a training set(n=130)and a validation set(n=57)using a ratio of 7:3.The fat attenuation index(FAI)model,three types of radiomics models[right coronary artery(RCA),left anterior descending artery(LAD),RCA combined with LAD(combined model)]were established by extracting and screening the radiomics features based on the proximal PCAT of RCA and LAD,and quantifying the FAI of the proximal RCA.The comprehensive model was developed using multifactorial logistic regression analysis to screen clinical and imaging data,combined with Rad-score of the combined model.Two classifiers,logistic regression(LR)and random forest(RF),were used to construct each of the above five models.The discriminative performance of the model was evaluated by receiver operating characteristic(ROC)curves,and the calibration curve was used to evaluate its calibration degree.Results The RCA combined with the LAD proximal PCAT screened 9 optimal radiomics features to calculate Rad-scores,among which glycated hemoglobin A1c(HbA1c),presence of high-risk plaques in the coronary tree(HRP)and Rad-scores were independent risk factors for diagnosing SA from UA.The RF model had higher discriminatory performance than the LR model and was statistically significant in the validation group(both P<0.05).Among the five models established with the classifier as RF,the area under the curve(AUC)of three radiomics models and the comprehensive model were higher than those of the FAI model(both P<0.05).However,there was no significant difference in AUC among the three radiomics models.The AUC of the comprehensive model training set(0.89)was higher than the AUC of the combined model(0.86,P=0.004).The calibration curve showed that three types of radiomics models and the comprehensive model were well calibrated.Conclusion The CCTA-based PCAT radiomic model is superior to the FAI model for diagnosing UA.Combining PCAT radiomics with clinical and imaging features(HbA1b,HRP)can further improve the diagnosis of UA.