Multicenter Observational Study of Multimodal AI for Upper GI Mesenchymal Tumor Diagnosis

Recruiting

• Conditions: Submucosal Tumor; Gastrointestinal Stromal Tumor (GIST); Leiomyoma; Schwannoma

• Registration Number: NCT07078136
• Lead Sponsor: Huazhong University of Science and Technology

Brief Summary

This study develops a multimodal AI model using endoscopic ultrasound, white-light endoscopy, and clinical information to support the diagnosis of upper GI mesenchymal tumors and the risk stratification of gastric GISTs.

Detailed Description

This is a multicenter, observational study designed to evaluate the diagnostic performance of a multimodal artificial intelligence (AI) model for the classification of upper gastrointestinal subepithelial lesions (SELs) and risk stratification of gastric gastrointestinal stromal tumors (gGISTs). The study combines retrospective image data for training and validation with prospectively recruited cases for testing.

Endoscopic ultrasound (EUS) images, white-light endoscopy (WLE) images, and relevant clinical data will be collected according to strict image quality control criteria. The multimodal AI model integrates these inputs using a multi-branch fusion strategy. A cross-validation trial will be conducted using prospectively recruited patients' data from multiple centers to compare the diagnostic and predictive performance of endoscopists with and without AI assistance for both lesion classification and risk stratification.

According to existing literature, no multimodal AI model has yet reported diagnostic performance for classifying SELs or for risk stratification of gastric gGISTs. It is assumed that the multimodal AI model will achieve a diagnostic accuracy of 95% for classifying upper gastrointestinal SELs and 95% for gGIST risk stratification. In comparison, the diagnostic accuracy of endoscopists is approximately 73.3%-75% for differentiating GIST from non-GIST and 72.4%-78.9% for risk stratification of gGISTs . GISTs account for about 67-68% of all lesions . Using a two-sided confidence interval with α = 0.05 and β = 0.2, and considering a 20% potential dropout rate, the minimum sample size required for prospective SEL classification is 65 cases, and 88 gGIST cases for risk stratification. Since the risk stratification task requires a larger sample size and GISTs are the common target of both tasks, the final planned sample size is 130 patients with upper GI SELs, which meets the statistical requirements for all primary endpoints.

The study team will screen patients based on the inclusion and exclusion criteria, ensure that all required examinations are completed to confirm eligibility, and record pre-treatment test results. All prospective participants will provide written informed consent before any study-related examinations.

This study is purely observational. No additional interventions will be performed on participants, nor will any additional costs be incurred. Patients' access to optimal diagnostic or treatment options will not be affected. The primary potential risk is the breach of patient privacy; the research team will establish a strict data security and monitoring plan and inform participants that their data will be used for clinical research purposes.

This study is purely observational. No additional interventions will be performed on participants, nor will any additional costs be incurred. Patients' access to optimal diagnostic or treatment options will not be affected. The primary potential risk is the breach of patient privacy; the research team will establish a strict data security and monitoring plan and inform participants that their data will be used for clinical research purposes.

Each enrolled participant will undergo diagnostic assessment by both the multimodal AI model and expert endoscopists. The AI model and expert interpretation will be blinded to each other. Final diagnosis will be confirmed by histopathology. Diagnostic performance will be compared using paired analysis. All statistical tests will be two-sided, and differences will be considered statistically significant at P \< 0.05. Continuous variables will be described as mean ± standard deviation. Categorical variables will be presented as counts and percentages. (1) Diagnostic Performance: The diagnostic performance of endoscopists and the AI model will be compared using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and area under the curve (AUC). F1-score (harmonic mean) and balanced accuracy will be calculated to address class imbalance (e.g., GIST vs. other lesions). (2) Continuous Data: Comparisons with baseline values will be conducted using paired t-tests, ANOVA, or rank-sum tests as appropriate. (3) Categorical Data: Group comparisons will use Chi-square tests (including CMH Chi-square test) or Fisher's exact test. (4) Baseline Comparability: Demographic and baseline characteristics will be compared using independent t-tests or Chi-square tests to assess group balance. (5) Effectiveness Analysis: The primary effectiveness endpoint is the diagnostic accuracy for GI subepithelial lesions. The difference in proportions and Youden index will be compared using the approximate normal Z test or Chi-square test with center effect control. (6) Software: All statistical analyses will be performed using SPSS version 26.0.

Study Timeline

• Status Verified: 2025-07
• Study First Submitted: 2025-07-02
• Study First Submitted QC: 2025-07-12
• Study First Posted: 2025-07-22
• Study Start: 2025-07-28
• Last Update Submitted: 2025-07-28
• Last Update Posted: 2025-07-31
• Primary Completion: 2026-03
• Study Completion: 2026-06-01

Recruitment & Eligibility

• Status: RECRUITING
• Sex: All
• Target Recruitment: 130

Inclusion Criteria

• Age ≥ 18 years old

• Patients with an upper gastrointestinal subepithelial lesion (SEL) identified by white-light endoscopy and who have completed an endoscopic ultrasound (EUS) examination

• Patients with a histopathological diagnosis of GIST confirmed by surgical or endoscopic resection, or other SELs confirmed by surgical resection, EUS-guided sampling, or other biopsy techniques

• EUS image quality meets the following quality control standards

  1. Equipment requirements: Olympus EU-ME2/ME1 processor (Olympus Medical Systems Corp., Tokyo, Japan); radial EUS scope (GF-UE260/GF-UE240; Olympus, Tokyo, Japan) or linear EUS scope (GF-UCT260/GF-UCT240; Olympus, Tokyo, Japan); miniature probe (UM2R/3R; Olympus, Tokyo, Japan); Pentax ARIETTA 850 processor (Pentax, Tokyo, Japan); radial EUS scope (EG-3670URK, Pentax, Tokyo, Japan); linear EUS scope (EG-3870UT, Pentax, Tokyo, Japan); Fujifilm SU-8000 or SU-9000 processor; linear EUS scope (EG-580UT, Fujifilm, Tokyo, Japan); radial EUS scope (EG-580UR, Fujifilm, Tokyo, Japan)
  2. EUS images clearly showing the lesion and surrounding tissue characteristics (at least 5 images or video); must include at least one image of the maximum lesion diameter, one image showing the layer of origin, and one image demonstrating the growth pattern (intraluminal/extraluminal)
  3. EUS images must not contain artificial annotations, such as measurement scales, biopsy needles, Doppler signals, or elastography overlays
  4. Image resolution must be at least 448 × 448 pixels

• WLE (white-light endoscopy) image quality meets the following standards: images must clearly show the lesion location, mucosal features, and margins; at least one close-up and one distant view

• Complete clinical data and histopathological reports must be available

Exclusion Criteria

• Age < 18 years old
• Absolute contraindications for EUS examination, history of gastric surgery, pregnancy, severe comorbidities, or known allergy to anesthetic agents
• EUS examination terminated prematurely due to esophageal stricture, obstruction, large space-occupying lesions, rapid changes in heart rate or respiratory rate, patient intolerance, or excessive residual food
• EUS image quality does not meet the required quality control standards
• Pathological specimens do not meet diagnostic requirements: insufficient biopsy tissue (only R0 resection specimens are accepted for the GIST group), or incomplete immunohistochemical staining (missing CD117/CD34/DOG-1 expression report for the GIST group)
• Pathological results indicate that the lesion is a metastatic tumor originating from another site

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Primary Outcome Measures

Name	Time	Method
Diagnostic accuracy of a multimodal AI model for differentiating gastrointestinal stromal tumors (GISTs) from other upper gastrointestinal mesenchymal tumors	After the training process of the multimodal AI model is completed，on average per year	Receiver operating characteristic (ROC) analyses, sensitivity, specificity, accuracy, positive predictive value and negative predictive value will be used to evaluate the efficacy of the model
Predictive accuracy of the multimodal AI model for risk stratification of GISTs	After the training process of the multimodal AI model is completed，on average per year	ROC analyses, sensitivity, specificity, accuracy, positive predictive value and negative predictive value will be used to evaluate the efficacy of the model

Secondary Outcome Measures

Name	Time	Method
Comparison of diagnostic accuracy between the multimodal AI model and experienced endoscopists for differentiating GISTs and non-GIST mesenchymal tumors	After the testing process of the multimodal AI model is completed，on average per year	The diagnostic accuracy of the AI model and expert endoscopists will be compared within the same participants, using histopathology as the gold standard. Receiver operating characteristic (ROC) analyses, sensitivity, specificity, accuracy, positive predictive value and negative predictive value will be used to evaluate the efficacy of the model
Comparison of Diagnostic Accuracy Between the Multimodal AI Model and Single-Modality Models	After the training process of the Multimodal AI model is completed，on average per year	Receiver operating characteristic (ROC) analyses, sensitivity, specificity, accuracy, positive predictive value and negative predictive value will be used to evaluate the efficacy of the model
Comparison of the predictive accuracy for GIST risk stratification between the multimodal AI model and experienced endoscopists	After the testing process of the multimodal AI model is completed，on average per year	The diagnostic accuracy of the AI model and expert endoscopists will be compared within the same participants, using histopathology as the gold standard. Receiver operating characteristic (ROC) analyses, sensitivity, specificity, accuracy, positive predictive value and negative predictive value will be used to evaluate the efficacy of the model

Trial Locations

Locations (1)

Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology; 🇨🇳 Wuhan, Hubei, China