NCT04607720

Clinical Research on a Novel Deep-learning Based System in Pancreatic Mass Diagnosis

Clinical Research on a Noval Deep-learning Based System in Benign and Malignant Pancreatic Masses Diagnosing Under Harmonic Contrast-enhanced Endoscopic Ultrasound

• Status: Unknown
• Phase: N/A
• Study type: Interventional
• Enrollment: 100 (estimated)
• Sponsor: The Third Xiangya Hospital of Central South University · Academic / Other
• Sex: All
• Age: 18 Years – 90 Years
• Healthy volunteers: Not accepted

Summary

In this study, a single-center, prospective, self-control, and blind design was adopted.

Detailed description

It is planned to enroll 100 patients with solid pancreatic masses who are to be diagnosed with contrast-enhanced harmonic endoscopy ultrasonography (CH-EUS) follow by EUS-guided fine-needle aspiration (EUS-FNA). First, all patients will be diagnosed both by the AI-assisted diagnosis system and the endoscopists under CH-EUS; Second, to compare the EUS-FNA under the guidance of the AI-assisted ultrasound diagnosis system and the EUS-FNA under manual identification, patients were randomly assigned to undergo EUS-FNA with or without the guidance of AI-assisted diagnosis system for the first two passes.

Conditions

• Solid Pancreatic Masses

Interventions

Type	Name	Description
PROCEDURE	artificial-EUS-FNA	The ultrasound endoscopists manually determined the region of interest (ROI) according to CH-EUS, and punctured 2 needles in the tumor area with a 22G needle, and each needle moved 15-20 times back and forth in the lesion. After puncture, insert the needle core, rinse the needle with 0.1ml sterile saline, and further apply air flushing to ensure that each tissue is sent to the slide.
PROCEDURE	AI-EUS-FNA	In the AI-assisted diagnosis system, the assisted targeting EUS-FNA unit is applied, and the ultrasound endoscopist selects the region of interest (ROI) according to the thermogram to perform targeted EUS-FNA puncture(2 needles in the same way).

Timeline

• Start date: 2020-01-01
• Primary completion: 2021-12-31
• Completion: 2022-06-30
• First posted: 2020-10-29
• Last updated: 2020-10-29

Locations

1 site across 1 country: China