Liver international : official journal of the International Association for the Study of the Liver · 2026
Radiomics-Based Assessment of Portal Hypertension Severity and Risk Stratification of Cirrhotic Patients Using Routine CT Scans.
СтатьяПоддержка исследования: вне правительства СШАНазвание и аннотация: Перевод пока не подготовлен
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.
Sin Celine, Watzenboeck Martin Luther, Iofinova Eugenia, Balcar Lorenz, Semmler Georg, Scheiner Bernhard, Lampichler Katharina, Mandorfer Mattias, Moga Lucile, Rautou Pierre-Emmanuel, Ronot Maxime, Menche Jörg, Reiberger Thomas, Scharitzer Martina
Аннотация
BACKGROUND & AIMS: To develop and validate a CT-based radiomics model to assess HVPG and predict a composite endpoint of liver-related events (LRE: decompensation and liver-related death) in patients with cirrhosis.
METHODS: This retrospective study included 357 cirrhosis patients, who received invasive HVPG measurements, 120 liver-healthy controls (training cohort) and 85 and 100 cirrhosis patients (internal and external validation cohorts, respectively), and contrast-enhanced abdominal CTs. After volumetric segmentation of the liver and spleen on CT, Bayesian parameter optimization was used for selection of extracted features and hyperparameter tuning in random forest or elastic net models. Prediction accuracy was evaluated using Pearson correlation coefficients of predicted ('radio-HVPG') and invasive HVPG. Discrimination between relevant HVPG cut-offs was determined by receiver operating characteristic (ROC) analysis. The predictive value of radio-HVPG and invasive-HVPG for LRE was compared using Cox regression models.
RESULTS: Radio-HVPG, predicted by an optimized random forest model based on 74 selected CT features, correlated with invasive-HVPG and detected clinically significant portal hypertension (CSPH: HVPG ≥ 10 mmHg) on the internal (Pearson r = 0.63, AUC 0.89 [95% CI: 0.81-0.96]) and external (Pearson r = 0.62, AUC 0.80 [95% CI: 0.64-0.91]) validation cohorts. Radio-HVPG predicted LRE when adjusting for MELD and albumin (adjusted HR: 1.14 [95% CI: 1.04-1.25], p = 0.005) and performed similarly to invasive-HVPG.
CONCLUSIONS: Radiomic features accurately predict HVPG in patients with cirrhosis and allow risk stratification for LRE in a radiomics-clinical signature.
METHODS: This retrospective study included 357 cirrhosis patients, who received invasive HVPG measurements, 120 liver-healthy controls (training cohort) and 85 and 100 cirrhosis patients (internal and external validation cohorts, respectively), and contrast-enhanced abdominal CTs. After volumetric segmentation of the liver and spleen on CT, Bayesian parameter optimization was used for selection of extracted features and hyperparameter tuning in random forest or elastic net models. Prediction accuracy was evaluated using Pearson correlation coefficients of predicted ('radio-HVPG') and invasive HVPG. Discrimination between relevant HVPG cut-offs was determined by receiver operating characteristic (ROC) analysis. The predictive value of radio-HVPG and invasive-HVPG for LRE was compared using Cox regression models.
RESULTS: Radio-HVPG, predicted by an optimized random forest model based on 74 selected CT features, correlated with invasive-HVPG and detected clinically significant portal hypertension (CSPH: HVPG ≥ 10 mmHg) on the internal (Pearson r = 0.63, AUC 0.89 [95% CI: 0.81-0.96]) and external (Pearson r = 0.62, AUC 0.80 [95% CI: 0.64-0.91]) validation cohorts. Radio-HVPG predicted LRE when adjusting for MELD and albumin (adjusted HR: 1.14 [95% CI: 1.04-1.25], p = 0.005) and performed similarly to invasive-HVPG.
CONCLUSIONS: Radiomic features accurately predict HVPG in patients with cirrhosis and allow risk stratification for LRE in a radiomics-clinical signature.