机器学习利用对比度增强超声波特征区分良性和恶性腮腺肿瘤

IF 2.3 3区医学 Q2 DENTISTRY, ORAL SURGERY & MEDICINE

Journal of Oral and Maxillofacial Surgery Pub Date : 2025-02-01 DOI:10.1016/j.joms.2024.10.018

Jie Shan MD , Yifei Yang MD , Hualian Liu PhD , Zhaoyao Sun PhD , Mingming Chen MD , Zhichao Zhu MD

{"title":"机器学习利用对比度增强超声波特征区分良性和恶性腮腺肿瘤","authors":"Jie Shan MD , Yifei Yang MD , Hualian Liu PhD , Zhaoyao Sun PhD , Mingming Chen MD , Zhichao Zhu MD","doi":"10.1016/j.joms.2024.10.018","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><div>Contrast-enhanced ultrasound (CEUS) is frequently used to distinguish benign parotid tumors (BPTs) from malignant parotid tumors (MPTs). Introducing machine learning may enable clinicians to preoperatively diagnose parotid tumors precisely.</div></div><div><h3>Purpose</h3><div>We aimed to estimate the diagnostic capability of machine learning in differentiating BPTs from MPTs.</div></div><div><h3>Study Design, Setting, and Sample</h3><div>A retrospective cohort study was conducted at the Third Affiliated Hospital of Soochow University. Patients who underwent parotidectomy and CEUS for untreated parotid tumors were included. Patients with recurrent tumors, inadequate specimens, or chemoradiotherapy were excluded.</div></div><div><h3>Predictor Variable</h3><div>Predictor variable was preoperative diagnosis coded as BPTs and MPTs based on the support vector machine (SVM) algorithms, laboratory, and CEUS variables.</div></div><div><h3>Main Outcome Variable(s)</h3><div>Outcome variable was pathological diagnosis coded as BPTs and MPTs.</div></div><div><h3>Covariates</h3><div>Covariate was demographics.</div></div><div><h3>Analyses</h3><div>A senior surgeon labeled patients' tumors as BPTs or MPTs, creating a clinical diagnosis. Patients were randomly divided into training (70%) and testing (30%) sets. After developing the SVM models using the training set, we evaluated their diagnostic performance on the testing set with the area under the receiver-operating characteristic curve (AUC), accuracy, positive predictive value, negative predictive value, sensitivity, and specificity. Delong's test was used to compare the AUC of SVM models, laboratory, and CEUS variables.</div></div><div><h3>Results</h3><div>The sample included 48 patients, and the testing set comprised 12 (25%) BPTs and 3 (6.25%) MPTs. Three CEUS variables (width, arrival time, and time to peak) and 3 laboratory variables (lymphocyte count, D-dimer, prognostic nutritional index) were identified through recursive feature elimination. Tested on the testing set, the SVM models with linear, polynomial, and radial kernels showed identical performance (AUC = 0.972, accuracy = 93.3%, positive predictive value = 75%, negative predictive value = 100%, sensitivity = 100%, specificity = 91.7%). They had larger AUC than SVM with sigmoid kernel (<em>P</em> = .18), width (<em>P</em> = .03), lymphocyte count (<em>P</em> = .02), D-dimer (<em>P</em> < .01), prognostic nutritional index (<em>P</em> = .03), arrival time (<em>P</em> = .02), time to peak (<em>P</em> = .04), CEUS diagnosis (<em>P</em> < .01), and clinical diagnosis (<em>P</em> < .01).</div></div><div><h3>Conclusion and Relevance</h3><div>The SVM algorithm differentiated BPTs from MPTs better than laboratory and CEUS variables.</div></div>","PeriodicalId":16612,"journal":{"name":"Journal of Oral and Maxillofacial Surgery","volume":"83 2","pages":"Pages 208-221"},"PeriodicalIF":2.3000,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Machine Learning Differentiates Between Benign and Malignant Parotid Tumors With Contrast-Enhanced Ultrasound Features\",\"authors\":\"Jie Shan MD , Yifei Yang MD , Hualian Liu PhD , Zhaoyao Sun PhD , Mingming Chen MD , Zhichao Zhu MD\",\"doi\":\"10.1016/j.joms.2024.10.018\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><h3>Background</h3><div>Contrast-enhanced ultrasound (CEUS) is frequently used to distinguish benign parotid tumors (BPTs) from malignant parotid tumors (MPTs). Introducing machine learning may enable clinicians to preoperatively diagnose parotid tumors precisely.</div></div><div><h3>Purpose</h3><div>We aimed to estimate the diagnostic capability of machine learning in differentiating BPTs from MPTs.</div></div><div><h3>Study Design, Setting, and Sample</h3><div>A retrospective cohort study was conducted at the Third Affiliated Hospital of Soochow University. Patients who underwent parotidectomy and CEUS for untreated parotid tumors were included. Patients with recurrent tumors, inadequate specimens, or chemoradiotherapy were excluded.</div></div><div><h3>Predictor Variable</h3><div>Predictor variable was preoperative diagnosis coded as BPTs and MPTs based on the support vector machine (SVM) algorithms, laboratory, and CEUS variables.</div></div><div><h3>Main Outcome Variable(s)</h3><div>Outcome variable was pathological diagnosis coded as BPTs and MPTs.</div></div><div><h3>Covariates</h3><div>Covariate was demographics.</div></div><div><h3>Analyses</h3><div>A senior surgeon labeled patients' tumors as BPTs or MPTs, creating a clinical diagnosis. Patients were randomly divided into training (70%) and testing (30%) sets. After developing the SVM models using the training set, we evaluated their diagnostic performance on the testing set with the area under the receiver-operating characteristic curve (AUC), accuracy, positive predictive value, negative predictive value, sensitivity, and specificity. Delong's test was used to compare the AUC of SVM models, laboratory, and CEUS variables.</div></div><div><h3>Results</h3><div>The sample included 48 patients, and the testing set comprised 12 (25%) BPTs and 3 (6.25%) MPTs. Three CEUS variables (width, arrival time, and time to peak) and 3 laboratory variables (lymphocyte count, D-dimer, prognostic nutritional index) were identified through recursive feature elimination. Tested on the testing set, the SVM models with linear, polynomial, and radial kernels showed identical performance (AUC = 0.972, accuracy = 93.3%, positive predictive value = 75%, negative predictive value = 100%, sensitivity = 100%, specificity = 91.7%). They had larger AUC than SVM with sigmoid kernel (<em>P</em> = .18), width (<em>P</em> = .03), lymphocyte count (<em>P</em> = .02), D-dimer (<em>P</em> < .01), prognostic nutritional index (<em>P</em> = .03), arrival time (<em>P</em> = .02), time to peak (<em>P</em> = .04), CEUS diagnosis (<em>P</em> < .01), and clinical diagnosis (<em>P</em> < .01).</div></div><div><h3>Conclusion and Relevance</h3><div>The SVM algorithm differentiated BPTs from MPTs better than laboratory and CEUS variables.</div></div>\",\"PeriodicalId\":16612,\"journal\":{\"name\":\"Journal of Oral and Maxillofacial Surgery\",\"volume\":\"83 2\",\"pages\":\"Pages 208-221\"},\"PeriodicalIF\":2.3000,\"publicationDate\":\"2025-02-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Oral and Maxillofacial Surgery\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0278239124009145\",\"RegionNum\":3,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"DENTISTRY, ORAL SURGERY & MEDICINE\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Oral and Maxillofacial Surgery","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0278239124009145","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"DENTISTRY, ORAL SURGERY & MEDICINE","Score":null,"Total":0}

引用次数: 0

摘要

背景：对比增强超声（CEUS）常用于区分良性腮腺肿瘤（BPT）和恶性腮腺肿瘤（MPT）。目的：我们旨在评估机器学习在区分良性腮腺肿瘤（BPT）和恶性腮腺肿瘤（MPT）方面的诊断能力：苏州大学附属第三医院开展了一项回顾性队列研究。研究纳入了接受腮腺切除术和CEUS治疗腮腺肿瘤的患者。不包括肿瘤复发、标本不足或接受化放疗的患者：预测变量是基于支持向量机（SVM）算法、实验室和CEUS变量编码为BPTs和MPTs的术前诊断：主要结果变量：结果变量为病理诊断，编码为 BPTs 和 MPTs：协变量：人口统计学：一位资深外科医生将患者的肿瘤标记为 BPTs 或 MPTs，从而做出临床诊断。患者被随机分为训练集（70%）和测试集（30%）。使用训练集建立 SVM 模型后，我们用接收者工作特征曲线下面积（AUC）、准确率、阳性预测值、阴性预测值、灵敏度和特异性评估了它们在测试集中的诊断性能。德隆检验用于比较 SVM 模型、实验室和 CEUS 变量的 AUC：样本包括 48 名患者，测试集包括 12 个（25%）BPT 和 3 个（6.25%）MPT。通过递归特征消除，确定了 3 个 CEUS 变量（宽度、到达时间和达到峰值的时间）和 3 个实验室变量（淋巴细胞计数、D-二聚体、预后营养指数）。在测试集上进行测试，具有线性、多项式和径向核的 SVM 模型显示出相同的性能（AUC = 0.972、准确率 = 93.3%、阳性预测值 = 75%、阴性预测值 = 100%、灵敏度 = 100%、特异性 = 91.7%）。与 SVM 相比，它们的 AUC 值更大，而 SVM 带有 sigmoid 核（P = .18）、宽度（P = .03）、淋巴细胞计数（P = .02）、D-二聚体（P 结论和相关性：SVM 算法比实验室和 CEUS 变量更能区分 BPT 和 MPT。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

Machine Learning Differentiates Between Benign and Malignant Parotid Tumors With Contrast-Enhanced Ultrasound Features

Background

Contrast-enhanced ultrasound (CEUS) is frequently used to distinguish benign parotid tumors (BPTs) from malignant parotid tumors (MPTs). Introducing machine learning may enable clinicians to preoperatively diagnose parotid tumors precisely.

Purpose

We aimed to estimate the diagnostic capability of machine learning in differentiating BPTs from MPTs.

Study Design, Setting, and Sample

A retrospective cohort study was conducted at the Third Affiliated Hospital of Soochow University. Patients who underwent parotidectomy and CEUS for untreated parotid tumors were included. Patients with recurrent tumors, inadequate specimens, or chemoradiotherapy were excluded.

Predictor Variable

Predictor variable was preoperative diagnosis coded as BPTs and MPTs based on the support vector machine (SVM) algorithms, laboratory, and CEUS variables.

Main Outcome Variable(s)

Outcome variable was pathological diagnosis coded as BPTs and MPTs.

Covariates

Covariate was demographics.

Analyses

A senior surgeon labeled patients' tumors as BPTs or MPTs, creating a clinical diagnosis. Patients were randomly divided into training (70%) and testing (30%) sets. After developing the SVM models using the training set, we evaluated their diagnostic performance on the testing set with the area under the receiver-operating characteristic curve (AUC), accuracy, positive predictive value, negative predictive value, sensitivity, and specificity. Delong's test was used to compare the AUC of SVM models, laboratory, and CEUS variables.

Results

The sample included 48 patients, and the testing set comprised 12 (25%) BPTs and 3 (6.25%) MPTs. Three CEUS variables (width, arrival time, and time to peak) and 3 laboratory variables (lymphocyte count, D-dimer, prognostic nutritional index) were identified through recursive feature elimination. Tested on the testing set, the SVM models with linear, polynomial, and radial kernels showed identical performance (AUC = 0.972, accuracy = 93.3%, positive predictive value = 75%, negative predictive value = 100%, sensitivity = 100%, specificity = 91.7%). They had larger AUC than SVM with sigmoid kernel (P = .18), width (P = .03), lymphocyte count (P = .02), D-dimer (P < .01), prognostic nutritional index (P = .03), arrival time (P = .02), time to peak (P = .04), CEUS diagnosis (P < .01), and clinical diagnosis (P < .01).

Conclusion and Relevance

The SVM algorithm differentiated BPTs from MPTs better than laboratory and CEUS variables.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

Journal of Oral and Maxillofacial Surgery 医学-牙科与口腔外科

CiteScore

4.00

自引率

5.30%

发文量

审稿时长

41 days

期刊介绍： This monthly journal offers comprehensive coverage of new techniques, important developments and innovative ideas in oral and maxillofacial surgery. Practice-applicable articles help develop the methods used to handle dentoalveolar surgery, facial injuries and deformities, TMJ disorders, oral cancer, jaw reconstruction, anesthesia and analgesia. The journal also includes specifics on new instruments and diagnostic equipment and modern therapeutic drugs and devices. Journal of Oral and Maxillofacial Surgery is recommended for first or priority subscription by the Dental Section of the Medical Library Association.