Potential value of CT-based comprehensive nomogram in predicting occult lymph node metastasis of esophageal squamous cell paralaryngeal nerves: a two-center study

• Published in: Journal of translational medicine 2024-04, Vol.22 (1), p.399-399, Article 399
• Main Authors: Xue, Ting, Wan, Xinyi, Zhou, Taohu, Zou, Qin, Ma, Chao, Chen, Jieqiong
• Format: Article
• Language: English
• Subjects: Adult; Aged; Algorithms; Analysis; Care and treatment; CT imaging; Data mining; Development and progression; Diagnosis; Diagnostic imaging; Esophageal cancer; Esophageal Neoplasms - diagnostic imaging; Esophageal Neoplasms - pathology; Esophageal squamous cell carcinoma; Esophageal Squamous Cell Carcinoma - diagnostic imaging; Esophageal Squamous Cell Carcinoma - pathology; Female; Health aspects; Humans; Laryngeal Nerves - diagnostic imaging; Laryngeal Nerves - pathology; Logistic Models; Lymph Nodes - diagnostic imaging; Lymph Nodes - pathology; Lymphatic Metastasis - pathology; Machine learning; Male; Medical imaging equipment; Medical research; Medicine, Experimental; Metastasis; Methods; Middle Aged; Multivariate Analysis; Nomogram; Nomograms; Para-laryngeal lymph node; Radiomics; Risk Factors; ROC Curve; Squamous cell carcinoma; Tomography, X-Ray Computed
• ISSN: 1479-5876; 1479-5876
• DOI: 10.1186/s12967-024-05217-4

The aim of this study is to construct a combined model that integrates radiomics, clinical risk factors and machine learning algorithms to predict para-laryngeal lymph node metastasis in esophageal squamous cell carcinoma. A retrospective study included 361 patients with esophageal squamous cell carcinoma from 2 centers. Radiomics features were extracted from the computed tomography scans. Logistic regression, k nearest neighbor, multilayer perceptron, light Gradient Boosting Machine, support vector machine, random forest algorithms were used to construct radiomics models. The receiver operating characteristic curve and The Hosmer-Lemeshow test were employed to select the better-performing model. Clinical risk factors were identified through univariate logistic regression analysis and multivariate logistic regression analysis and utilized to develop a clinical model. A combined model was then created by merging radiomics and clinical risk factors. The performance of the models was evaluated using ROC curve analysis, and the clinical value of the models was assessed using decision curve analysis. A total of 1024 radiomics features were extracted. Among the radiomics models, the KNN model demonstrated the optimal diagnostic capabilities and accuracy, with an area under the curve (AUC) of 0.84 in the training cohort and 0.62 in the internal test cohort. Furthermore, the combined model exhibited an AUC of 0.97 in the training cohort and 0.86 in the internal test cohort. A clinical-radiomics integrated nomogram can predict occult para-laryngeal lymph node metastasis in esophageal squamous cell carcinoma and provide guidance for personalized treatment.