Oral radiology · 2026
Positron emission tomography/computed tomography radiomics-based prediction of cervical nodal metastasis in patients with tongue squamous cell carcinoma.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Nakamura Shin, Kuribayashi Ami, Oki Arisa, Watanabe Hiroshi, Imaizumi Akiko, Taguchi Miharu, Miura Masahiko
Аннотация
OBJECTIVES: To develop and validate a logistic regression model that integrates clinical and fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT)-derived radiomic features to predict cervical lymph node metastasis in tongue squamous cell carcinoma.
METHODS: This retrospective study included 316 patients who underwent preoperative FDG-PET/CT between December 2013 and December 2023, classified into analysis (n = 250) and validation (n = 66) cohorts. Primary tumours were segmented using standardised uptake value (SUV)-based thresholds (≥ 40% of SUVmax). Metabolic parameters such as SUVmax and 107 radiomic features were extracted. Significant variables in the univariate analysis were analysed using multivariable logistic regression. Model performance was assessed using receiver operating characteristic analysis, with the predicted probability (PP) cut-off optimised using Youden's index. Diagnostic performance was compared with lymph node SUVmax and expert visual assessments.
RESULTS: Four independent predictors-clinical growth pattern, SUVpeak, grey level co-occurrence matrix inverse difference moment normalised, and grey level size zone matrix low grey-level zone emphasis-were included in the final model. In the analysis cohort, the model achieved an area under the curve (AUC) of 0.88, sensitivity of 0.84, and specificity of 0.79 at an optimal PP cut-off of 0.34. AUCs for lymph node SUVmax and expert visual interpretation were 0.76 and 0.79, respectively. In external validation, accuracy was 0.80 versus 0.68 for visual assessment.
CONCLUSIONS: A transparent logistic regression model integrating clinical growth patterns and FDG-PET/CT metabolic and radiomic features demonstrated high discrimination and maintained favourable accuracy in external validation, supporting its potential use in individualised preoperative risk stratification.
METHODS: This retrospective study included 316 patients who underwent preoperative FDG-PET/CT between December 2013 and December 2023, classified into analysis (n = 250) and validation (n = 66) cohorts. Primary tumours were segmented using standardised uptake value (SUV)-based thresholds (≥ 40% of SUVmax). Metabolic parameters such as SUVmax and 107 radiomic features were extracted. Significant variables in the univariate analysis were analysed using multivariable logistic regression. Model performance was assessed using receiver operating characteristic analysis, with the predicted probability (PP) cut-off optimised using Youden's index. Diagnostic performance was compared with lymph node SUVmax and expert visual assessments.
RESULTS: Four independent predictors-clinical growth pattern, SUVpeak, grey level co-occurrence matrix inverse difference moment normalised, and grey level size zone matrix low grey-level zone emphasis-were included in the final model. In the analysis cohort, the model achieved an area under the curve (AUC) of 0.88, sensitivity of 0.84, and specificity of 0.79 at an optimal PP cut-off of 0.34. AUCs for lymph node SUVmax and expert visual interpretation were 0.76 and 0.79, respectively. In external validation, accuracy was 0.80 versus 0.68 for visual assessment.
CONCLUSIONS: A transparent logistic regression model integrating clinical growth patterns and FDG-PET/CT metabolic and radiomic features demonstrated high discrimination and maintained favourable accuracy in external validation, supporting its potential use in individualised preoperative risk stratification.