Skip to content

Application and validation of generative artificial intelligence virtual enhanced CT in the diagnosis and treatment of pancreatic tumors: a multicenter, retrospective diagnostic trial

Application and validation of generative artificial intelligence virtual enhanced CT in the diagnosis and treatment of pancreatic tumors: a multicenter, retrospective diagnostic trial

Status
Recruiting
Phases
Unknown
Study type
Observational
Source
ChiCTR
Registry ID
ChiCTR2600119243
Enrollment
Unknown
Registered
2026-02-25
Start date
2026-02-25
Completion date
Unknown
Last updated
2026-03-02

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

Conditions

Pancreatic tumors

Interventions

Gold Standard:The histopathological results of surgically resected specimens or needle biopsy are used as the final diagnosis
Index test:Virtual Enhanced CT Generation System Based on Conditional Diffusion Models

Sponsors

Shanghai Changhai Hospital
Lead Sponsor

Eligibility

Sex/Gender
All
Age
18 Years to 100 Years

Inclusion criteria

Inclusion criteria: 1. Age >= 18 years; 2. Pancreatic tumour group: pathologically confirmed pancreatic tumour (PDAC or other tumour); 3. normal control group: initial imaging reported as normal pancreas and no progression of pancreatic lesions on >=6 months imaging follow-up (CT or MRI) during the retrospective collection 2015-2025; 4. have complete plain, arterial phase and portal phase enhanced CT data from the same examination; 5. Layer thickness of =5mm and complete coverage of the pancreas and surrounding key vessels; 6. For the subset of resectability assessment: detailed surgical records and postoperative pathology reports must be available.

Exclusion criteria

Exclusion criteria: 1. Those who have a history of pancreatic treatment or major upper abdominal surgery that has significantly changed their anatomical structure; 2. Severe metal artifacts or motion artifacts in the image, resulting in unreadable anatomy; 3. The real enhancement CT reinforcement effect is not good and cannot be used as the training truth value; 4. The registration of the flat scan and enhanced images fails, and paired training data cannot be constructed.

Design outcomes

Primary

MeasureTime frame
Sensitivity;Specificity;Area under the ROC curve;

Secondary

MeasureTime frame
Structural similarity;Peak signal-to-noise ratio;

Countries

China

Contacts

Public ContactKai Cao

Shanghai Changhai Hospital

mdkaicao163@163.com+86 185 0211 3743

Outcome results

None listed

Source: ChiCTR (via WHO ICTRP) · Data processed: Mar 14, 2026