单发肺部小结节良恶性诊断的预测模型-现代医学

单发肺部小结节良恶性诊断的预测模型

单位：南通大学附属海安医院影像科, 江苏海安 226600

分类号：R734.2

出版年·卷·期（页码）：2022·50·第一期（35-40）

摘要：

目的： 通过构建单发肺部小结节良恶性质诊断的预测模型，以提高临床早期鉴别肺部恶性结节的准确性。方法：回顾性总结2017年4月至2020年4月入我院判断为单发肺部小结节患者112例(结节直径8 mm~3 cm)，经64层螺旋CT增强扫描和重建处理，良恶性经CT引导穿刺活检或者手术病理证实，其中恶性结节40例(恶性组)，良性结节72例(良性组)。采用单因素分析比较两组患者性别、年龄、结节位置、结节直径、结节内部和外部纹理特征，其中内部特征包括分叶、空洞、钙化、血供，外部纹理特征包括毛刺、不规则、支气管征和血管束。多因素Logistic回归分析筛选影响性质判断的危险因素，绘制列线图预测模型、受试者工作特征曲线(ROC)，计算模型预测的准确性，Hosmer-Lemeshow 检验计算预测效能。结果： 恶性组患者的年龄比良性组增加，结节内部和外部纹理特征数量增多(P＜0.05)，但两组患者性别、结节位置、结节直径比较差异不明显(P＞0.05)。Logistic回归分析筛选发现，患者年龄、结节直径、结节内部和外部纹理特征数量是影响性质判断的独立危险因素(P＜0.05)。列线图预测模型显示，患者年龄和结节直径越大，结节内部和外部纹理特征数量越多，提示恶性结节的可能性就越高。ROC分析显示，该模型预测的准确性为0.896；Hosmer-Lemeshow 检验计算预测效能为91.07%，拟合效果良好。结论：经64层螺旋CT增强扫描和重建处理单发肺部小结节，构建恶性结节性质判断的预测模型主要指标包括患者年龄、结节直径、结节内部和外部纹理特征数量，有较好的准确性和应用价值。

Objective:To improve the accuracy of early clinical diagnosis of malignant single lung nodules by constructing a predictive model for the diagnosis of benign and malignant single lung nodule.Methods:A retrospective summary of 112 patients who were admitted to our hospital from April 2017 to April 2020 and judged to be a single pulmonary nodule(nodule diameter 8 mm-3 cm), after 64-slice spiral CT enhanced scanning and reconstruction processing, confirmed by CT-guided needle biopsy or surgical pathology,confirmed including 40 cases of malignant nodules(malignant group) and 72 cases of benign nodules(benign group). Single factor analysis was used to compare the gender, age, nodule location, nodule diameter, internal and external texture features of the nodule, the internal features including lobes, cavities, calcification, blood supply, and external texture features including burrs, irregularities, bronchial signs and vascular bundles. Multi-factor Logistic regression analysis screened out the risk factors that affected the nature of judgment. manomogram prediction model showed that the greater the patient age and nodule diameter, the greater the number of nodule internal and external markings, the higher the likelihood of malignant nodules.The ROC analysis showed that the prediction accuracy of this model was 0.896, and the Hosmer-Lemeshow test was 91.07%, with good fit.Results:The age of the patients in malignant group was higher than that in benign group, and the number of internal and external texture features ofnodules increased(P<0.05), but there was no significant difference in gender, nodule location and nodular diameter between the two groups(P>0.05). Logistic regression analysis showed thatpatient's age,nodule diameter and the number of internal and external texture features ofnodules were independent risk factors that affecting the judgment of nature(P<0.05). Nomogram prediction model shows that the larger the patient's age andnodule diameter, the more the number of internal and external texture features of the nodule, suggesting the higher the probability ofmalignant nodule. ROC analysis shows that the prediction accuracy of the model is 0.896; the prediction efficiency of Hosmer-Lemeshow testis 91.07%, and the fitting effect is good.Conclusion:After 64-slice spiral CT enhanced scanning and reconstruction, a single small lung nodule has been processed to construct a predictive model for judging the nature of malignant nodules. The main indicators include patient age, nodule diameter, and the number of internal and external texture features of the nodule with good accuracy and application value.

参考文献：

[1] 徐国厚，黄海峡，陈斌，等.某体检中心23695例体检者首次胸部低剂量CT筛查结果及肺结节相关易感因素的研究[J].复旦学报(医学版)，2020，9(4)：1-7.
[2] MAC M H,NAIDICH D P,GOO J M,et al.Guidelines for management of incidental pulmonary nodules detected on CT images:from the fleischner society 2017[J].Radiology,2017,284(1):228-243.
[3] LIU Y,BALAGURUNATHAN Y,ATWATER T,et al.Radiological image traits predictive of cancer status in pulmonary nodules[J].Clin Cancer Res,2017,23(6):1442-1449.
[4] 王婧璇，林岚，赵思远，等.基于深度学习的肺结节计算机断层扫描影像检测与分类的研究进展[J].生物医学工程学杂志，2019，36(4)：670-676.
[5] 陈淑娇，谢丽萍，陈懿.肺内微小结节CT影像征象与病理表现在早期肺癌诊断中的临床应用[J].中华肺部疾病杂志(电子版)，2020，13(4)：521-523.
[6] 胡磊，汪志亮，汪涛，等.孤立性肺结节CT诊断与鉴别诊断[J].中国CT和MRI杂志，2020，18(8)：39-42.
[7] CHI J N,ZHANG S,YU X S,et al.A novel pulmonary nodule detection model based on multi-step cascaded networks[J].Sensors(Basel),2020,20(15):4301.
[8] 范丽，方梦捷，董迪，等.影像组学对磨玻璃结节型肺腺癌病理亚型的预测效能[J].中华放射学杂志，2017，51(12)：912-917.
[9] 张晓菊.《肺结节诊治中国专家共识(2018版)》解读[J].中华实用诊断与治疗杂志，2019，33(1)：1-3.
[10] MAZZONE P J,SILVESTRI G A,PATEL S,et al.Screening for lung cancer:chest guideline and expert panel report[J].Chest,2018,153(4):954-985.
[11] 冯光，贺延法，董建立.CT纹理特征和形态征象对肺部纯磨玻璃结节侵袭性的诊断价值[J].中国医疗设备，2020，35(7)：77-80.
[12] KIM H，LEE D，CHO W S，等.基于CT的深度学习模型在术前呈亚实性结节中鉴别侵袭性肺腺癌:基于体积logistic模型的诊断效能与放射科医生比较[J].国际医学放射学杂志，2020，43(4)：501-502.
[13] 张静，武志峰，鄂林宁，等.肺结节(≤2 cm)及其周围组织的影像组学特征在其良恶性鉴别中的价值[J].中国临床医学影像杂志，2020，31(7)：478-482.
[14] ZHAO G,YU X,CHEN W,et al.Computed tomography-guided preoperative semi-rigid hook-wire localization of small pulmonary nodules:74 cases report[J].J Cardiothorac Surg,2019,14(1):149.
[15] 鄂林宁，张娜，王荣华，等.计算机体层摄影术纹理分析对孤立性肺结节良恶性鉴别诊断的价值[J].中华肿瘤杂志，2018，40(11)：847-850.
[16] ONISHI Y,TERAMOTO A,TSUJIMOTO M,et al.Automated pulmonary nodule classification in computed tomography images using a deep convolutional neural network trained by generative adversarial networks[J].Biomed Res Int,2019,2(1):154-156.
[17] 顾春东.周围型小肺癌诊疗问题[J].中国医师进修杂志，2019，42(7)：577-579.
[18] 杨尚文，胡安宁，徐亚运，等.CT图像分辨率对人工智能肺结节辅助诊断系统诊断准确性的影响[J].医学影像学杂志，2020，30(6)：965-968.
[19] 何义昆.CT纹理分析对孤立性肺结节的诊断价值探讨[J].基层医学论坛，2020，24(19)：2756-2757.
[20] BEIG N,KHORRAMI M,ALILOU M,et al.Perinodular and intranodular radiomic features on lung CT images distinguish adenocarcinomas from granulomas[J].Radiology,2019,290(3):783-792.

服务与反馈：

【文章下载】【发表评论】【查看评论】【加入收藏】

提示：您还未登录，请登录！点此登录