MedPath

Development and Prospective Validation of a Digital Pathology-based Artificial Intelligence Diagnostic Model for Pan-cancer Lymphatic Metastasis

Recruiting
Conditions
Cancer
Lymphatic Metastasis
Interventions
Diagnostic Test: Artificial intelligence (AI)-based diagnostic model
Registration Number
NCT06517979
Lead Sponsor
Sun Yat-Sen Memorial Hospital of Sun Yat-Sen University
Brief Summary

The goal of this diagnostic test is to develop an artificial intelligence (AI)-based pan-cancer universal diagnostic model for detecting pathological lymph node metastasis (LNM), and prospectively evaluate its apllication value in the real-world clinical practice.

Investigators will compare the diagnostic performance (sensitivity, specificity, etc.) of the AI model and routine pathological report issued by pathologists, to see if the AI model can improve the clinical workflow of pathological evaluation of cancer LNM in in the real world.

Detailed Description

Lymph node metastasis (LNM) is a common mode of cancer metastasis, and accurate postoperative pathological lymph node staging is of great significance for further treatment and prognosis assessment. However, the current pathological evaluation of lymph nodes relies on manual examination by pathologists, which has a relatively low diagnostic efficiency and is prone to missed-diagnosis for micro metastatic lesions.

Therefore, investigators are to develope an artificial intelligence (AI)-based diagnostic model for detecting pathological cancer lymph node metastasis based on deep learning algorithms, and evaluate its apllication value in the real-world clinical settings.

This study is a diagnostic test with no intervention measures, planning to collect pathological slides of formalin-fixed, paraffin-embedded lymph nodes resected from the enrolled patients and digitise them into whole-slide images (WSIs). The AI model will analyse the WSIs and generate pixel-level heatmaps and slide-level diagnostic results (with or without LNM). The routine pathological examination will be performed as usual. These two processes will not interfere with each other. And if there are inconsistency in slide-level classification between AI and routine pathological examination, investigators would convene senior pathologists for discussion to make the final decision (immunohistochemistry would be performed if necessary). The final result will be presented to the patient in the form of a pathological report.

Recruitment & Eligibility

Status
RECRUITING
Sex
All
Target Recruitment
10000
Inclusion Criteria
  • Patients with cancer, undergoing radical tumor resection and lymph node dissection.
  • Patients with complete clinical and pathological information.
Exclusion Criteria
  • The patient refused to participate in this diagnostic test.

Study & Design

Study Type
OBSERVATIONAL
Study Design
Not specified
Arm && Interventions
GroupInterventionDescription
Patients with cancer undergoing LNDArtificial intelligence (AI)-based diagnostic modelPatients undergo radical tumor resection and lymph node dissection (LND)
Primary Outcome Measures
NameTimeMethod
sensitivityFor each enrolled patient, the diagnosis results of AI model will be obtained in servel days after lymph node dissection, and the sensitivity of the AI model will be evaluated through study completion, an average of 3 year.

the number of correctly diagnosed positive slides (with lymphatic metastasis), to be divided by the number of positive slides in total

Secondary Outcome Measures
NameTimeMethod
specificityFor each enrolled patient, the diagnosis results of AI model will be obtained in servel days after lymph node dissection, and the sensitivity of the AI model will be evaluated through study completion, an average of 3 year.

the number of correctly diagnosed negative slides (without lymphatic metastasis), to be divided by the number of negative slides in total

Trial Locations

Locations (1)

Sun Yat-sen Memorial Hospital of Sun Yat-sen University

🇨🇳

Guangzhou, Guangdong, China

© Copyright 2025. All Rights Reserved by MedPath