An In Silico Trial to Evaluate Prospectively the Performance of a Radiomics Algorithm for UIP Compared to Medical Doctors
Study Details
Study Description
Brief Summary
The purpose of this study is to compare AI performance to doctor's performance in the evaluation of IPF/UIP and ILDs without UIP(proven by biopsy).
Condition or Disease | Intervention/Treatment | Phase |
---|---|---|
|
Study Design
Arms and Interventions
Arm | Intervention/Treatment |
---|---|
IPF/UIP_CT based patients with an ILD and a pathological UIP pattern and a final diagnosis of IPF |
Other: Radiomics model to classify between IPF with UIP pattern and ILDs without UIP pattern
The aim is to evaluate the performance of AI with the performance of doctors
|
IPF/UIP_Biopsy based patients with a final diagnosis of IPF but a less typical HRCT pattern( lung biopsy required for the diagnosis) |
Other: Radiomics model to classify between IPF with UIP pattern and ILDs without UIP pattern
The aim is to evaluate the performance of AI with the performance of doctors
|
ILD but not IPF and prove by biopsy not UIP patients with an ILD and a pathological non-UIP pattern |
Other: Radiomics model to classify between IPF with UIP pattern and ILDs without UIP pattern
The aim is to evaluate the performance of AI with the performance of doctors
|
Normal Normal healthy patients |
Other: Radiomics model to classify between IPF with UIP pattern and ILDs without UIP pattern
The aim is to evaluate the performance of AI with the performance of doctors
|
Outcome Measures
Primary Outcome Measures
- The performance of Radiomics algorithm compared to the ground truth [May 2021]
Reporting the performance measure: accuracy
Secondary Outcome Measures
- Comparing the performance of the radiomics algorithm to that of physicians [June 2021]
Correctness of the diagnosis - the most probable thin-section pattern (dichotomous outcome: yes or no)
Other Outcome Measures
- Comparing the performance of radiomics algorithm to that of individual doctor [June 2021]
Correctness of the diagnosis - the most probable thin-section pattern (dichotomous outcome: yes or no)
- Comparing the reproducibility of diagnosis between the doctors [June 2021]
Measuring the concordance in diagnosis in between the physicians, and for each physician separately, using the Concordance correlation coefficient.
Eligibility Criteria
Criteria
Inclusion Criteria:
- the availability of non-contrast-enhanced HRCT
Exclusion Criteria:
-
the use of contrast enhancement
-
images containing metal or motion artifacts
-
Images reconstructed with a slice thickness larger than 1.5 mm
Contacts and Locations
Locations
Site | City | State | Country | Postal Code | |
---|---|---|---|---|---|
1 | Maastricht University | Maastricht | Limburg | Netherlands | 6229ER |
Sponsors and Collaborators
- Maastricht University
Investigators
None specified.Study Documents (Full-Text)
None provided.More Information
Publications
None provided.- ISTRU