Explainable multimodal AI for diagnosing early gastric cancer from endoscopic images

Development and Validation of an Explainable Artificial Intelligence Model for Early Gastric Cancer Diagnosis Using Multimodal Endoscopic Imaging

Observational The First Affiliated Hospital of Soochow University · NCT07551466

This project will try an explainable AI that combines white-light and image-enhanced endoscopy pictures plus clinical data to help detect early gastric cancer in adults with suspicious gastric lesions who underwent endoscopic submucosal dissection.

Quick facts

Study typeObservational
Enrollment100 (estimated)
Ages18 Years to 80 Years
SexAll
SponsorThe First Affiliated Hospital of Soochow University Academic / other
Locations1 site (Suzhou)
Trial IDNCT07551466 on ClinicalTrials.gov

What this trial studies

This observational project will collect endoscopic images (white-light and image-enhanced/NBI) and clinical data from adults who underwent endoscopic submucosal dissection for suspicious gastric lesions. A lesion detection model will first identify regions of interest, then quantitative image features (color, texture, morphology, mucosal structure) and deep-learning–derived features will be extracted from both imaging modalities. These imaging features will be integrated with clinical variables to build a multimodal, explainable prediction model. Model performance will be tested on held-out data and reported using AUROC, sensitivity, specificity, accuracy, and calibration metrics.

Who should consider this trial

Good fit: Adults (≥18 years) with suspicious gastric lesions who had both white-light imaging and magnifying endoscopy with narrow-band imaging and subsequently underwent endoscopic submucosal dissection with available histopathology are ideal candidates.

Not a fit: Patients without adequate WLI or ME‑NBI imaging, those who did not undergo ESD, or those with non-adenocarcinoma histologies or deeply invasive lesions are unlikely to benefit from this model.

Why it matters

Potential benefit: If successful, the model could improve accuracy of early gastric cancer detection and give interpretable decision support to endoscopists, potentially enabling earlier and more appropriate treatment.

How similar studies have performed: Previous AI models have improved endoscopic detection of gastric neoplasia, but most relied on single-modality images and lacked interpretability, while multimodal explainable approaches remain relatively novel.

Eligibility criteria

Show full inclusion / exclusion criteria
Inclusion Criteria:

* Age ≥18 years
* Suspicious gastric lesions identified on white-light imaging (WLI)
* Preoperative biopsy indicating precancerous lesions (dysplasia or intraepithelial neoplasia) or adenocarcinoma, with preoperative magnifying endoscopy with narrow-band imaging (ME-NBI) performed
* Patients meeting the absolute indications for endoscopic submucosal dissection (ESD) and who underwent ESD

Exclusion Criteria:

* Non-adenocarcinoma histological types (e.g., lymphoma)
* Patients who did not undergo ME-NBI examination or did not receive ESD
* Lesions invading the muscularis propria or deeper layers
* Missing or indeterminate postoperative histopathological results

Where this trial is running

Suzhou

Study contacts

How to participate

  1. Review the eligibility criteria above with your treating physician.
  2. Visit the official trial page on ClinicalTrials.gov for the most current contact information and recruitment status.
  3. Contact the listed study coordinator or principal investigator to request pre-screening. Pre-screening is free and never obligates you to enroll.
Conditions Early Gastric CancerArtificial IntelligenceMultimodal ImagingEndoscopic ImagingExplainable AIDeep Learning
Last reviewed 2026-06-13 by the Find a Trial editorial team. Information on this page is for educational purposes and is not medical advice. Always consult qualified healthcare professionals about clinical trial participation.