简介 颈部肥胖与健康和非健康个体的心血管风险有关。我们的目的是评估从计算机断层扫描 (CT) 检查中提取的颈部解剖测量值作为心血管疾病及其危险因素的预测指标的效用。方法 我们调查了 2013 年至 2020 年间在两家医院接受静脉造影 CT 颈部检查的患者。排除患有颈部恶性肿瘤、既往接受过颈部手术、年龄 <18 岁、人口统计信息不完整和图像质量不合格的患者。我们对颈部解剖结构进行了 18 次独立测量，这些测量与心血管危险因素和疾病以及相关实验室值和药物相关。所有多变量线性回归均针对性别和 BMI 进行控制。 p<0.05 的关联被认为具有统计显着性。然后使用随机森林（一种非线性预测算法）将测量结果用于预测高血压。结果 2013 年至 2020 年间，进行了约 20,000 次颈部增强 CT 检查。应用纳入标准后，458 名患者仍留在研究人群中。八项测量（所有测量都包括颈部肥胖的一个组成部分）显示解剖测量和心血管危险因素之间存在统计学上显着的关联。与 CT 测量值增加最常相关的危险因素是 2 型糖尿病。因此，接受胰岛素治疗的患者所有八项测量的平均值显着较高。之前被诊断患有高脂血症的人和正在接受高血压药物治疗的人的测量值也显着增加。随机森林预测算法的接受者操作特征 (AUROC) 值下面积为 0.68，这意味着我们的测量结果可以很好地预测高血压疾病状态。结论颈部CT检查提取的脂肪组织测量值与高血压、糖尿病和高脂血症等心血管危险因素相关。颈部解剖测量的机器学习模型可以潜在地识别有心血管疾病风险的患者。版权所有 © 2024，Lopez 等人。

Introduction Neck adiposity has been related to cardiovascular risk in healthy and nonhealthy individuals. Our objective was to evaluate the utility of anatomic neck measurements extracted from computed tomography (CT) examinations as a predictor of cardiovascular disease and its risk factors. Methods We investigated patients who had a CT neck examination with intravenous contrast performed at two hospitals between 2013 and 2020. Patients with a neck malignancy, prior neck surgery, age <18 years, incomplete demographic information, and inadequate image quality were excluded. We performed 18 separate measurements of neck anatomy which were correlated with cardiovascular risk factors and disease, as well as relevant lab values and medications. All multivariable linear regressions were controlled for gender and BMI. Associations with p<0.05 were considered statistically significant. The measurements were then used to predict hypertension using random forest, a non-linear prediction algorithm. Results Approximately 20,000 neck CT examinations with contrast were performed between 2013-2020. After applying the inclusion criteria, 458 patients remained in the study population. Eight measurements (all of which include a component of neck adiposity) showed a statistically significant association between anatomic measurements and cardiovascular risk factors. The risk factor most often associated with increases in CT measurements was type 2 diabetes. Accordingly, patients on insulin treatment had a significantly higher average for all eight measurements. Significant measurement increases were also found in those previously diagnosed with hyperlipidemia and in those being treated with hypertension medications. The area under the receiver operating characteristic (AUROC) value of the random forest prediction algorithm was 0.68, meaning our measurements were a good predictor of hypertensive disease status. Conclusion Adipose tissue measurements extracted from CT examinations of the neck are associated with cardiovascular risk factors including hypertension, diabetes, and hyperlipidemia. Machine learning models of anatomic neck measurements can potentially identify patients at risk for cardiovascular disease.Copyright © 2024, Lopez et al.