Skip to content

Construction and application of a deep learning-based computer-assisted endoscopy model for determining the depth of invasion in digestive cancer

Construction and application of deep learning-based computer aided endoscopic ultrasonography model for determining the depth of invasion in gastrointestinal cancer

Status
Recruiting
Phases
Unknown
Study type
Observational
Source
ChiCTR
Registry ID
ChiCTR2300070637
Enrollment
Unknown
Registered
2023-04-19
Start date
2022-08-01
Completion date
Unknown
Last updated
2023-06-04

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

early gastrointestinal cancer

Interventions

Gold Standard:Pathological staging

Sponsors

Fudan University Shanghai Cancer Center
Lead Sponsor

Eligibility

Sex/Gender
All
Age
18 Years to 99 Years

Inclusion criteria

Inclusion criteria: Training Sets:1) Patients with gastrointestinal cancer (including esophagus, stomach and rectum) confirmed by endoscopic mucosal dissection (ESD) or surgery during January 2017 to December 2022 in each center; 2) Complete reports of pathological diagnosis and pathological staging (TNM); 3) The patient underwent endoscopic ultrasonography before surgery, and the medical electronic system had clear endoscopic ultrasonography and white light endoscopy images with clear tumor information. Test Sets:1) Patients with gastrointestinal cancer (including esophagus, stomach and rectum) confirmed by endoscopic mucosal dissection (ESD) or surgery during June 2022 to June 2024 in each center; 2) Complete reports of pathological diagnosis and pathological staging (TNM); 3) The patient underwent endoscopic ultrasonography before surgery, and the medical electronic system had clear endoscopic ultrasonography and white light endoscopy images with clear tumor information.

Exclusion criteria

Exclusion criteria: 1) Other types of malignant tumors (such as malignant mesenchymal tumors, etc.); 2) The endoscopic image does not contain tumor information or the image model is not clear, making it difficult to identify the tumor levels.

Design outcomes

Primary

MeasureTime frame
the model's diagnostic performance;

Countries

China

Contacts

Public ContactHe yiping

Fudan University Shanghai Cancer Center

hyp21@126.com+86 18121299009

Outcome results

None listed

Source: ChiCTR (via WHO ICTRP) · Data processed: Feb 4, 2026