Abdominal radiology (New York) · 2025
Magnetic resonance imaging radiomics for predicting hepatocellular carcinoma recurrence following resection or ablation: a systematic review and meta-analysis.
СтатьяСистематический обзорМетаанализНазвание 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.
Maung Soe Thiha, Tangkijvanich Pisit, Chaiteerakij Roongruedee
Аннотация
OBJECTIVE: We assessed predictive performance of MRI radiomics models for hepatocellular carcinoma (HCC) recurrence after curative treatments, aiming to inform future research on optimizing MRI-based radiomics for HCC management.
METHODS: This study followed PRISMA and we systematically searched for studies predicting HCC recurrence and microvascular invasion (MVI). Quality assessment used Quality Assessment of Diagnostic Accuracy Studies, Radiomics Quality Score, and METhodological RadiomICs Score (METRICS) tools. Meta-analyses were performed with random-effects models, assessing heterogeneity.
RESULTS: Radiomics models achieved a pooled concordance index (C-index) of 0.78(95%CI:0.71-0.85) in training and 0.73(95%CI:0.69-0.77) in validation cohorts, outperforming clinico-radiological models. For recurrence prediction after liver resection, radiomics models achieved pooled sensitivity/specificity of 0.81(0.74-0.87)/0.85(0.77-0.90) in training (AUC: 0.87; accuracy: 0.81[95%CI:0.76-0.86]) and 0.80(0.73-0.86)/0.74(0.68-0.80) in validation cohorts (AUC: 0.83; accuracy: 0.79[0.75-0.83]). For thermal ablation, sensitivity/specificity were 0.80(0.60-0.90)/0.79(0.74-0.84) in training (AUC: 0.79; accuracy: 0.78[0.72-0.84]) and 0.79(0.69-0.87)/0.78(0.57-0.90) in validation (AUC:0.82; accuracy:0.76[0.67-0.84]). For MVI prediction, radiomics models achieved 0.87(0.78-0.92)/0.88(0.80-0.93) in training (AUC: 0.93; accuracy: 0.86[0.80-0.91]), and 0.88(0.80-0.94)/0.79(0.72-0.84) in validation cohorts (AUC: 0.87; accuracy: 0.82[0.77-0.86]). Subgroup analysis revealed significantly improved specificity for models using PyRadiomics and LASSO for feature extraction and selection, respectively.
CONCLUSION: MRI radiomics models outperformed routine clinico-radiological models, with potentially enhanced clinical decision-making. Future studies should focus on cost-effectiveness and external validation to support broader clinical integration.
ADVANCES IN KNOWLEDGE: This study demonstrates that MRI radiomics models significantly enhance predictive accuracy for HCC recurrence and MVI, highlighting their potential as noninvasive tools to guide personalized treatment planning.
METHODS: This study followed PRISMA and we systematically searched for studies predicting HCC recurrence and microvascular invasion (MVI). Quality assessment used Quality Assessment of Diagnostic Accuracy Studies, Radiomics Quality Score, and METhodological RadiomICs Score (METRICS) tools. Meta-analyses were performed with random-effects models, assessing heterogeneity.
RESULTS: Radiomics models achieved a pooled concordance index (C-index) of 0.78(95%CI:0.71-0.85) in training and 0.73(95%CI:0.69-0.77) in validation cohorts, outperforming clinico-radiological models. For recurrence prediction after liver resection, radiomics models achieved pooled sensitivity/specificity of 0.81(0.74-0.87)/0.85(0.77-0.90) in training (AUC: 0.87; accuracy: 0.81[95%CI:0.76-0.86]) and 0.80(0.73-0.86)/0.74(0.68-0.80) in validation cohorts (AUC: 0.83; accuracy: 0.79[0.75-0.83]). For thermal ablation, sensitivity/specificity were 0.80(0.60-0.90)/0.79(0.74-0.84) in training (AUC: 0.79; accuracy: 0.78[0.72-0.84]) and 0.79(0.69-0.87)/0.78(0.57-0.90) in validation (AUC:0.82; accuracy:0.76[0.67-0.84]). For MVI prediction, radiomics models achieved 0.87(0.78-0.92)/0.88(0.80-0.93) in training (AUC: 0.93; accuracy: 0.86[0.80-0.91]), and 0.88(0.80-0.94)/0.79(0.72-0.84) in validation cohorts (AUC: 0.87; accuracy: 0.82[0.77-0.86]). Subgroup analysis revealed significantly improved specificity for models using PyRadiomics and LASSO for feature extraction and selection, respectively.
CONCLUSION: MRI radiomics models outperformed routine clinico-radiological models, with potentially enhanced clinical decision-making. Future studies should focus on cost-effectiveness and external validation to support broader clinical integration.
ADVANCES IN KNOWLEDGE: This study demonstrates that MRI radiomics models significantly enhance predictive accuracy for HCC recurrence and MVI, highlighting their potential as noninvasive tools to guide personalized treatment planning.