The purpose of this clinical study is to conduct a multi-center, big data study to create a neural network decision model for predicting treatment efficacy and prognosis based on multi-modal, multi-temporal imaging features combined with tumor microenvironment scores. It will also use various model interpretation techniques to clarify the role and mechanism of key biomarkers or strongly associated biomarker groups in treatment efficacy and prognosis. Ultimately, it aims to achieve the research and application of AI treatment strategies combining multi-modal imaging and digital pathology to guide clinicians in the personalized treatment strategies for patients with esophageal squamous cell carcinoma.

Inclusion Criteria:

1. Aged 18-70 years;
2. Histologically confirmed esophageal carcinoma by biopsy;
3. No prior antitumor therapy received.

Exclusion Criteria:

1. Contraindications to MRI examination;
2. Poor compliance with antitumor therapy;
3. Unwillingness to participate in the study;
4. Image quality inadequate for diagnostic requirements.

This study will enroll 7,000 esophageal carcinoma patients from multiple centers. Participants will receive CT and mpMRI scans before, during, and after treatment, as required for evaluating surgery, neoadjuvant therapy, or definitive chemoradiation. PET-CT will be performed if distant metastasis is suspected. Whole-slide digital scanning will be applied to biopsy and surgical specimens for pathological analysis.