AI-assisted Endoscopic Ultrasound Grading of Early Esophageal Cancer Invasion Depth: A Multicenter, Prospective, Randomized Cohort Study

AI-assisted Endoscopic Ultrasound Grading of Early Esophageal Cancer Invasion Depth: A Multicenter, Prospective, Randomized Cohort Study

This study mainly uses an artificial intelligence system to assist in the classification of the depth of invasion of early esophageal squamous cell carcinoma under ultrasound endoscopy, providing a basis for preoperative T staging and diagnosis and treatment decisions.

For patients with early esophageal squamous cell carcinoma and precancerous lesions who met the inclusion and exclusion criteria and voluntarily participated in this project, they were randomly divided into the AI group and the conventional group by central randomization, with 100 cases in each group（anticipated）. Randomization method: The personnel responsible for randomization at the center (who do not participate in the inclusion of subjects) log in to the central randomization system to obtain a randomization number, and finally form a randomization allocation table. Blinding implementation: The observation group and control group determined on the random allocation table were marked as A and B respectively, and then the operating physician implemented protocol A or B. Main indicators: Grading judgment of infiltration depth, pathological consistency

Inclusion Criteria:

• Satisfy ①⑧⑨ and one of the following conditions simultaneously: ②③④⑤⑥⑦ ① Age over 18 years old, ② Esophageal ulcer, ③ low-grade intraepithelial neoplasia, ④ high-grade intraepithelial neoplasia, ⑤ patients with esophageal squamous cell carcinoma, ⑥ white patches of esophageal mucosa, ⑦ esophageal polyps, ⑧ with endoscopic examination records and detailed pathological records, ⑨ agree to participate in the study;

Exclusion Criteria:

• ① Patients who have undergone esophageal cancer surgery, ② those with a history of radiotherapy and chemotherapy for esophageal cancer, ③ patients with missing data.