| 摘要: |
| 摘 要 目的 基于超声影像组学、超声图像特征及临床资料构建列线图模型，探讨其术前预测甲状腺乳头状癌（PTC）患者颈侧区淋巴结（LNLN）转移的临床价值。方法 选取经手术病理证实为PTC患者161例，按7:3比例随机分为训练集112例和验证集49例，均有完整的超声检查及临床资料，并根据病理结果分为LNLN转移阳性组50例和LNLN转移阴性组111例（已改）。基于训练集的灰阶超声图像勾画并提取感兴趣区（ROI）的影像组学特征，采用最小绝对收缩和选择算子（LASSO）算法筛选与PTC患者LNLN转移相关的特征，计算每个病灶的影像组学分数（RS）。采用Logistic回归分析筛选临床资料、超声图像特征及超声影像组学特征，分别构建临床模型、超声图像特征模型、超声影像组学模型和联合模型。绘制受试者工作特征（ROC）曲线分析各模型预测PTC LNLN转移的效能。应用校准曲线评估各模型的校准度。结果 单因素和多因素Logistic回归分析显示，性别（OR=1.117）和瘤体最大直径（OR=2.935）为LNLN转移的独立影响因素，基于上述2个独立影响因素分别构建临床模型和超声图像特征模型。经LASSO回归降维共筛选出6个系数非零的超声影像组学特征用于构建超声影像组学模型。训练集中LNLN转移阳性组和阴性组的RS分别为（0.51±0.25）分、（0.22±0.19）分；验证集中LNLN转移阳性组和阴性组的RS分别为（0.68±0.28）分、（0.44±0.23）分，差异均有统计学意义（均P＜0.05）。ROC曲线分析显示，在训练集和验证集中，临床模型预测LNLN转移阳性组和阴性组的曲线下面积（AUC）分别为0.635和0.538，超声图像特征模型的AUC分别为0.757和0.741，超声影像组学模型的AUC分别为0.824和0.747，联合模型的AUC分别为0.843和0.778；超声影像组学模型和联合模型的AUC均高于临床模型，差异有统计学意义（P＜0.05）。校准曲线显示，超声影像组学模型和联合模型的校准度均较高，与实际结果的一致性均较好。结论 基于超声影像组学、超声图像特征及临床资料构建列线图模型术前预测PTC患者LNLN转移有重要的临床价值。 |
| 关键词: 超声检查 影像组学 甲状腺乳头状癌 颈侧区淋巴结 |
| DOI： |
| 投稿时间：2024-05-10修订日期：2024-08-27 |
| 基金项目:无锡市“双百”中青年医疗卫生后备拔尖人才（HB2023001）；无锡市科协软科学研究课题（KX-23-B071） |
|
| Ultrasound-Based Radiomic Nomogram to Predict The Lateral Neck Lymph Node Metastasis of Papillary Thyroid Carcinoma |
| chensichen,dingyan |
| () |
| Abstract: |
| ABSTRACT Objective To construct a column-line diagram model based on ultrasound imaging histology, ultrasound image features and clinical data, and to explore its clinical value for preoperative prediction of lymph node metastasis in the lateral neck region (LNLN) in patients with papillary thyroid carcinoma (PTC). Methods 161 patients with PTC confirmed by surgical pathology were selected and randomly divided into 112 cases in the training set and 49 cases in the validation set according to the ratio of 7:3, all of whom had complete ultrasonographic and clinical data and were divided into 50 cases in the LNLN metastasis-positive group and 111 cases in the LNLN metastasis-negative group according to the pathological results. Based on the gray-scale ultrasound images of the training set to outline and extract the imaging histology features in the region of interest (ROI), the least absolute shrinkage and selection operator (LASSO) algorithm was used to screen the features related to LNLN metastasis in patients with PTC, and the imaging histology score (RS) of each lesion was calculated. Logistic regression analysis was used to screen the clinical data, ultrasound image features and ultrasound imaging histology features, and the clinical model, ultrasound image features model, ultrasound imaging histology model and joint model were constructed, respectively. The efficacy of each model in predicting PTC LNLN transfer was analyzed by plotting the subject operating characteristic (ROC) curve. Calibration curves were applied to assess the calibration of each model.Results Univariate and multivariate logistic regression analyses showed that gender (OR=1.117) and tumor maximum diameter (OR=2.935) were independent influences on LNLN metastasis, and clinical models and ultrasound image feature models were constructed based on the above 2 independent influences, respectively. A total of 6 ultrasound image histologic features with non-zero coefficients were screened by LASSO regression downscaling for the construction of the ultrasound image histologic model. The RS of the LNLN metastasis-positive and negative groups in the training set were (0.51±0.25) and (0.22±0.19), respectively; and that of the LNLN metastasis-positive and negative groups in the validation set were (0.68±0.28) and (0.44±0.23), respectively, and the differences were statistically significant (both P<0.05).The analysis of the ROC curves showed that the clinical model predicted the LNLN metastasis-positive and negative groups in the training and validation sets. set and validation set, the area under the curve (AUC) of the clinical model for predicting LNLN metastasis in the positive and negative groups were 0.635 and 0.538, respectively, and the AUCs of the ultrasound image characterization model were 0.757 and 0.741, respectively, the AUCs of the ultrasound imaging histology model were 0.824 and 0.747, respectively, and the AUCs of the combined model were 0.843 and 0.778, respectively; The AUCs of the ultrasound imaging histology model and the combined model were higher than those of the clinical model, and the differences were statistically significant (P < 0.05). The calibration curves showed that both the ultrasound imaging histology model and the joint model were highly calibrated, and both were in good agreement with the actual results.Conclusion Constructing a nomogram model based on ultrasound imaging histology, ultrasound image features and clinical data for preoperative prediction of LNLN metastasis in PTC patients has important clinical value. |
| Key words: Ultrasonography Radiomics Papillary thyroid carcinoma Lateral neck lymph node |
