Effect of Blink-Based Training on Cancer Detection in Colorectal Polyp Images by Medical Students/Non GI-trainees (Blink)

Assessment of Blink Impression for Cancer Detection in Colorectal Polyps Among Medical Students and Non-GI Trainees: A Pre-Post Intervention Study

This study investigates whether a brief educational intervention using Blink features can improve medical students' and non-GI trainees' ability to detect colorectal cancer in static polyp images. Secondary aims include evaluating changes in specificity, confidence, and interobserver agreement, determining which Blink features support accurate detection, and examining the link between the number of features recognized and diagnostic performance. The study will recruit medical students and non-GI trainees without prior training in polyp morphology or endoscopic image interpretation, who will complete an online pre- and post-intervention image-based survey.

Study Overview

• Status: Completed
• Conditions: Colorectal Neoplasms; Colorectal Polyps
• Intervention / Treatment: Behavioral: Blink Features Training Video

Detailed Description

Background and Rationale:

Colorectal cancer (CRC) remains a leading cause of cancer-related mortality in Western countries, though it is largely preventable by detecting and removing precursor lesions such as colorectal polyps. While most polyps are small and benign, 1-2% are large (≥20 mm) non-pedunculated colorectal polyps (LNPCPs), which carry a markedly higher risk of invasive cancer (reported rates 6-15%, depending on morphology, histology, and location).

Accurate optical diagnosis of cancer in LNPCPs is critical for guiding treatment strategy-piecemeal endoscopic resection, en bloc resection, or surgery. However, endoscopists often underperform in identifying cancer within these lesions. Studies have reported correct cancer identification rates as low as 20-40%, even among trained endoscopists, contributing to unnecessary surgeries for benign polyps and missed diagnoses in malignant cases.

One contributing factor is the complexity of existing classification systems, which are rarely applied consistently in clinical practice. Simplified tools may improve accuracy and applicability.

The Blink framework, inspired by Malcolm Gladwell's concept of rapid, intuitive decision-making and aligned with Kahneman's System 1 thinking, condenses cancer recognition into six easily observed features of LNPCPs:

Fold deformation Extra redness Chicken skin mucosa Depression Spontaneous bleeding Ulceration

These Blink features can be recognized without advanced imaging and provide a structured, intuitive framework for rapid cancer detection. Previous research has shown that teaching these features improves endoscopists' diagnostic sensitivity. Building on this, the present study evaluates whether a brief Blink-based intervention can improve cancer detection among medical students and non-GI trainees with no prior training in polyp morphology.

Primary Objective:

To assess whether a short educational intervention (2-minute training video on Blink features) improves the sensitivity of medical students and non-GI trainees in detecting cancer in colorectal polyps using static images.

Secondary Objectives

• To evaluate changes in specificity, self-reported confidence, and interobserver agreement before and after the intervention.
• To identify which Blink features are associated with accurate cancer detection.
• To assess the relationship between the number of Blink features identified and diagnostic accuracy.

Study Design:

Design: Prospective interventional study with pre- and post-intervention assessments.

Setting: Online survey distributed to medical students and non-GI trainees affiliated with the Vrije Universiteit Brussel.

Intervention: 2-minute video training on the six Blink features, followed by re-assessment of images.

Target Population:

Medical students and non-GI trainees affiliated with the Vrije Universiteit Brussel without prior endoscopy experience.

Sample size: 50-100 participants (yielding 1,000-2,000 individual image evaluations).

• Study Type: Interventional
• Enrollment (Actual): 65
• Phase: Not Applicable

Contacts and Locations

Study Locations

• Belgium
  • Brussels Capital
    • Jette, Brussels Capital, Belgium, 1090
      • UZ Brussel

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: Yes

Description

Inclusion Criteria:

• Images of large non-pedunculated colorectal polyps (LNPCPs) obtained from patients who have provided informed consent for the anonymous use of polyp images during colonoscopy (EC-2024-200).
• Medical students and non-GI trainees who have provided written informed consent to participate in the study.

Exclusion Criteria:

• Participants (students or non-GI trainees) with prior experience in endoscopy or polyp detection, or those who do not provide informed consent.
• Images deemed to be of insufficient quality by the principal investigator.
• Patients who do not provide informed consent for image use and data collection.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Diagnostic
• Allocation: N/A
• Interventional Model: Single Group Assignment
• Masking: None (Open Label)

Number of Arms

1

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: Single Arm: Blink Features Training; Medical students and non-GI trainees complete a baseline assessment of colorectal polyp images, receive a brief 2-minute educational video on Blink features, and then complete a post-intervention assessment.	Behavioral: Blink Features Training Video; A brief (2-minute) educational video introducing six Blink features (fold deformation, extra redness, chicken skin mucosa, depression, spontaneous bleeding, ulceration) to improve recognition of colorectal cancer in large non-pedunculated colorectal polyps.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Change in sensitivity for detection of colorectal cancer in polyp images	Sensitivity will be calculated as the proportion of correctly identified cancerous polyps out of all cancerous polyps presented. Comparison will be made between pre-intervention and post-intervention assessments.	Immediately before and after the educational intervention (within one online survey session, approximately 15-20 minutes).

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Change in specificity for cancer detection	Specificity: proportion of cancerous polyps correctly identified as cancerous. Compared pre- vs post-intervention using the same participant as their own control.	Immediately before and after the educational intervention (single online session, ~15-20 minutes).
Change in self-reported diagnostic confidence	Mean confidence score per case (e.g., 1-5 Likert scale) averaged per participant, compared pre- vs post-intervention.	Immediately before and after the educational intervention (same session).
Change in interobserver agreement (Fleiss' kappa)	Agreement among participants on cancer vs non-cancer classification, computed as Fleiss' kappa and compared pre- vs post-intervention	Immediately before and after the educational intervention (same session).
Correlation between presence of individual Blink features and correct cancer classification	For each endoscopic image, participants will indicate the presence or absence of predefined Blink features (fold deformation, extra redness, chicken skin mucosa, depression, spontaneous bleeding, ulceration). Each feature will be analyzed separately. The presence of each feature will be correlated with diagnostic accuracy, defined as the proportion (%) of correct classifications (cancer vs non-cancer) across participants, using histology as the reference standard. Measurement Tool: Online survey with image annotation for Blink features and classification task for cancer vs non-cancer. Unit of Measure: Correlation coefficient (r) between presence/absence of each Blink feature and diagnostic accuracy (%).	Baseline (pre-intervention) and immediately after the educational intervention (within the same online survey session, ~15-20 minutes).
Correlation between number of Blink features identified and diagnostic accuracy	For each image, participants will indicate the presence or absence of predefined Blink features (fold deformation, extra redness, chicken skin mucosa, depression, spontaneous bleeding, ulceration). The per-image/per-participant count of features identified will then be correlated with diagnostic accuracy (% of correct cancer vs non-cancer classifications). Measurement Tool: Online survey with image annotation for Blink features and classification task for cancer vs non-cancer (histology as reference standard). Unit of Measure: Correlation coefficient (r) between number of Blink features identified and diagnostic accuracy (%).	Immediately after the educational intervention (within the same online survey session, ~15-20 minutes).

Collaborators and Investigators

• Sponsor: Universitair Ziekenhuis Brussel

Study record dates

Study Major Dates

• Study Start (Actual): December 20, 2025
• Primary Completion (Actual): January 27, 2026
• Study Completion (Actual): January 27, 2026

Study Registration Dates

• First Submitted: September 8, 2025
• First Submitted That Met QC Criteria: February 5, 2026
• First Posted (Actual): February 10, 2026

Study Record Updates

• Last Update Posted (Actual): February 10, 2026
• Last Update Submitted That Met QC Criteria: February 5, 2026
• Last Verified: February 1, 2026

More Information

Terms related to this study

• Keywords: Colorectal Polyps; Optical Diagnosis; Blink
• Additional Relevant MeSH Terms: Neoplasms by Site; Neoplasms; Intestinal Diseases; Gastrointestinal Neoplasms; Digestive System Neoplasms; Digestive System Diseases; Gastrointestinal Diseases; Intestinal Neoplasms; Rectal Diseases; Colonic Diseases; Colorectal Neoplasms
• Other Study ID Numbers: 25311_Blink (Other Identifier: UZ Brussel)

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES
• IPD Plan Description: De-identified individual participant data (IPD) underlying the reported results will be made available upon reasonable request from qualified researchers after publication of the study results. Data will be shared directly by the principal investigator following a data access agreement
• IPD Sharing Time Frame: De-identified IPD and supporting information will be available starting 6 months after publication of the primary results and will remain available for 5 years.
• IPD Sharing Access Criteria: Data will be shared with qualified researchers upon reasonable request and after signing a data access agreement.
• IPD Sharing Supporting Information Type: STUDY_PROTOCOL; SAP

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No