Project

Optimising the Human Detection of Cancer within Colorectal Polyps with a Simple Clinical Algorithm

Code
T004723N
Duration
01 October 2023 → 30 September 2027
Funding
Research Foundation - Flanders (FWO)
Research disciplines
  • Natural sciences
    • Machine learning and decision making
  • Medical and health sciences
    • Gastro-enterology
Keywords
Colorectal polyps Colorectal cancer Endoscopic diagnosis
 
Project description

Population
Colorectal cancer is a leading cause of cancer death in the Western world. It is preventable by
colonoscopy and the detection and removal of colorectal polyps (polypectomy). Discrimination of
polyps at high risk of containing cancer is of critical importance to determine whether polypectomy is
appropriate and to avoid adverse patient outcomes associated with incorrect treatment selection.
Intervention
Using evidence-based features of polyps that determine high risk for cancer we created a simple
online algorithm that may help endoscopists determine the risk of cancer within polyps.
It was validated (vs microscopy and expert opinion) in a videobased study of 37 endoscopists. A
moderate (kappa 0.54) interobserver agreement and high (97.5%) negative predictive value for
cancer within polyps was obtained and the tool determined correct treatment in 71.3% of cases.
User feedback was positive regarding usability and applicability.
To maximise the impact of the algorithm on patient outcomes, further video validation, validation
during live endoscopy, validation pre- and post- a standardized learning intervention, development of
the algorithm into a user friendly format and dissemination via professional societies is proposed
during this project.
Comparison
Existing tools to determine the presence of cancer within colorectal polyps are complex and not user
friendly. This leads to poor performance of endoscopists at discriminating polyps which contain
cancer.
Outcome
The consequences for patients can be significant if cancer within a polyp is missed or if polyps are
incorrectly labelled as containing cancer e.g. polypectomy without the chance of cancer cure,
incomplete polypectomy and unnecessary use of surgery (which has much higher risks to the patient
than polypectomy) for polyps without cancer. The widespread adoption of a validated algorithm to
accurately determine the risk of cancer within a colorectal polyp offers the chance to avoid these
risks.