Colorectal Cancer, Colorectal Polyp
Conditions
Keywords
Artificial intelligence, Computer-aided diagnosis, Optical diagnosis, Colorectal polyps, Blue light imaging
Brief summary
Primary, this study aims to develop and validate a computer-aided diagnosis (CADx) system for the characterization of colorectal polyps. Second, this study evaluates the effect of using a clinical classification model Blue Light Imaging Adenoma Serrated International (BASIC) on the diagnostic accuracy of the optical diagnosis of colorectal polyps compared to intuitive optical diagnosis for both expert endoscopists and novices.
Detailed description
Optical diagnosis of colorectal polyps, the in-vivo characterization of the histology by endoscopists, is of increasing interest for clinical endoscopy practice. Recent studies have shown that thresholds for optical diagnosis are met in highly selected groups of expert endoscopists, but the same is not true in community endoscopy practices. In order to improve optical diagnosis, imaging enhancement techniques and the use of artificial intelligence are proposed. This observational study developes a computer-aided diagnosis (CADx) system to differentiate between benign and (pre-)malignant CRPs, using state-of-the-art machine learning methods and deep learning architectures. For the development, HDWL and BLI images are used. The CADx is trained using histology as gold standard. The CADx is externally validated using on a set of 60 colorectal polyps. This study will evaluate if the optical diagnosis of colorectal polyps can be improved with the aid of CADx. In addition, both expert endoscopists and novices optically diagnose the colorectal polyps. In the first, pre-training phase, endoscopists optically diagnose colorectal polyps based on intuition. Afterwards, in the post-training phase, the same set of colorectal polyps is optically diagnosed based on a clinical classification system; BLI Adenoma Serrated International Classification (BASIC). This study will evaluate if the optical diagnosis of colorectal polyps can be improved with the aid of BASIC in both expert and non-expert hands.
Interventions
OTHERComputer-aided diagnosis (CADx)
Optical diagnosis of colorectal polyps made with computer-aided diagnosis (CADx) using state-of-the-art machine learning methods and deep learning architectures.
OTHERBLI Adenoma Serrated International Classification (BASIC)
Optical diagnosis of colorectal polyps made with BLI Adenoma Serrated International Classification (BASIC), both by exert endoscopists and novices.
Sponsors
Catharina Ziekenhuis Eindhoven
Eindhoven University of Technology
Maastricht University Medical Center
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
* Patients with at least one colorectal polyp found and resected during colonoscopy in regular care; * Availability of at least one high definition white light image and one Blue Light Imaging (BLI) image of the colorectal polyp; * Overall high quality of the colorectal polyp image; * Availability of the histological results of the colorectal polyp; * Minimal age of 18 years old.
Exclusion criteria
* Objection to participate in medical scientific research, reported in the medical file; * Endoscopic instruments visible at the colorectal polyp image; * More than one polyp visible at the colorectal polyp image.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Diagnostic accuracy of CADx versus endoscopists | 6 months | The diagnostic accuracy of characterizing colorectal polyps (into benign versus (pre-)malignant) made by CADx in comparison to the diagnostic accuracy of both expert endoscopists and novices. In which histology is the gold standard. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Diagnostic metrics of CADx | 6 months | The sensitivity, specificity, negative and positive predictive value of characterizing colorectal polyps (into benign versus (pre-)malignant) made by CADx. |
| AUC of CADx | 6 months | Area Under ROC Curve (AUC) of CADx for characterizing colorectal polyps (into benign versus (pre-)malignant) made by CADx. |
| Diagnostic metrics of endoscopists | 6 months | The sensitivity, specificity, negative and positive predictive value of characterizing colorectal polyps (into hyperplastic polyp, adenoma, sessile serrated adenoma, or adenocarcinoma) of both expert endoscopists and novices. |
| Diagnostic accuracy of BASIC versus intuition | 6 months | The diagnostic accuracy of both expert endoscopists and novices in characterizing colorectal polyps (into hyperplastic polyp, adenoma, sessile serrated adenoma, or adenocarcinoma) based on the clinical classification model BASIC, in comparison to the diagnostic accuracy based on intuition. In which histology is the gold standard. |
| Interobserver agreement | 6 months | The interobserver agreement of experts, novices and CADx for characterizing colorectal polyps. |
| Computation time of CADx | 6 months | The computation time per image of CADx for characterizing colorectal polyps. |
| Diagnostic metrics of endoscopists for high confidence diagnosis | 6 months | The diagnostic accuracy, sensitivity, specificity, negative and positive predictive value of both expert endoscopists and novices for optical diagnosis made with high (\>90%) confidence. |
Countries
Netherlands
Outcome results
None listed