Clinical Scorecard: Classification of Polyp Characteristics by NICE for Colonoscopy Screening
At a Glance
Category
Detail
Condition
Colorectal cancer and precancerous polyps
Key Mechanisms
NICE classification categorizes polyps by vessels, surface patterns, and colour using Narrow-Band Imaging during colonoscopy
Target Population
Patients undergoing colonoscopy screening for colorectal cancer
Care Setting
Endoscopy units performing colonoscopy with Narrow-Band Imaging
Key Highlights
NICE classification divides polyps into three types: type-1 (hyperplastic), type-2 (adenomas), and type-3 (deep submucosal invasive cancer).
Accurate on-site polyp classification using NBI and NICE reduces costs and time compared to histology but depends on clinician experience.
Deep learning models, specifically a ResNet-101 architecture, can classify polyp features (colour, vessels, surface pattern) aligned with NICE criteria to support clinical decision-making.
Guideline-Based Recommendations
Diagnosis
Use colonoscopy with Narrow-Band Imaging to identify and classify polyps based on NICE criteria.
Histological analysis remains the gold standard but is resource-intensive; on-site classification can guide immediate management.
Management
Remove adenomatous polyps (type-2) to reduce colorectal cancer risk.
Hyperplastic polyps (type-1) may often be left untreated.
Recognize deep invasive cancers (type-3) for appropriate oncologic referral.
Monitoring & Follow-up
Maintain high adenoma detection rates during colonoscopy to reduce cancer risk.
Use balanced datasets and quality imaging to improve classification accuracy.
Risks
Misclassification may lead to unnecessary polypectomy or missed adenomas, impacting patient outcomes.
Low-quality imaging frames can impair classification; use 'indistinguishable' label to manage uncertain cases.
Patient & Prescribing Data
Patients undergoing colonoscopy screening with detected polyps
Accurate classification of polyp features supports targeted polypectomy, reducing colorectal cancer risk and avoiding unnecessary procedures.
Clinical Best Practices
Employ Narrow-Band Imaging during colonoscopy to enhance visualization of polyp features.
Use NICE classification to guide real-time decision-making on polyp management.
Incorporate deep learning tools trained on clinically interpretable features to assist clinicians, especially less experienced endoscopists.
Label low-quality or unclear images as 'indistinguishable' to avoid misclassification.
Balance adenoma and non-adenoma cases in training datasets to improve model performance.
