Computed tomography- and magnetic resonance imaging-based multi-modality radiomics models for predicting survival in esophageal cancer patients
Multi-modality CT and MRI radiomics models integrating clinical factors significantly outperform single-modality models in predicting disease-free and overall survival for esophageal squamous cell carcinoma patients.
Published Feb 5, 2026Paper ↗
Only vote on papers you've read. Sign in to vote.
Abstract
OBJECTIVES: This study explores the integration of radiomics features from contrast-enhanced CT and MRI with clinical and radiological risk factors for optimal prognosis models of disease-free survival (DFS) and overall survival (OS) in esophageal squamous cell carcinoma (ESCC). MATERIALS AND METHODS: A retrospective study is undertaken of 371 ESCC patients who underwent contrast-enhanced CT and MRI with StarVIBE sequence, from September 2014 to December 2019. Prognosis models for DFS and OS were developed using cross-validation and Elastic-Net-Cox regression. Patients were grouped by treatment type (surgery, chemoradiotherapy, neoadjuvant therapy) to create single-modality and multi-modality models (Model-S, Model-CRT, and Model-nT). Model performance was evaluated using nomograms, calibration curves, and decision curves. RESULTS: Twelve optimal prognosis models were identified. For DFS, MRI c-indices were 0.595, 0.608, and 0.721, while CT c-indices were 0.686, 0.616, and 0.667. For OS, MRI c-indices were 0.692, 0.597, and 0.650, and CT c-indices were 0.656, 0.623, and 0.695. Multi-modality models demonstrated c-indices of 0.750, 0.695, 0.839 for DFS and 0.898, 0.777, 0.819 for OS. CONCLUSION: Single-modality radiomics models exhibit limited predictive ability for DFS and OS in ESCC patients, whereas multi-modality radiomics models enhance predictive accuracy significantly.