When a doctor performs a colonoscopy and finds a colorectal polyp, one of the most important pieces of information is how large it is. Polyp size directly influences whether a patient needs follow-up surveillance sooner or later, and which removal technique the doctor should use.
Despite this importance, measuring polyp size accurately is harder than it sounds. Studies show that doctors visually estimating size during a colonoscopy are only correct about 48 to 65 percent of the time, and they tend to overestimate how big a polyp is.
A reliable, validated measurement method is needed, especially as artificial intelligence tools are being developed to help doctors size polyps automatically. For AI to work well, it needs to be trained on accurate measurements. This study set out to formally test whether a simple tool called a digital caliper meets that standard.
In research and AI development, a reference standard is the most accurate measurement available, the one that other methods are compared against to see how well they perform. Without a trusted reference standard, it is impossible to know whether a new tool or algorithm is actually accurate.
Previously, high-resolution digital microscopy was shown to be an excellent reference standard for measuring polyp size. Digital microscopes can photograph removed polyp specimens and measure them with great precision. However, these microscopes require specialized equipment, careful setup, and trained staff, making them difficult to use in large studies or everyday clinical settings.
Caliper-based measurements, where a small digital measuring tool is applied directly to the removed polyp specimen right in the endoscopy room, are far simpler and cheaper. Many ongoing AI studies already use calipers as their reference, but until this study, no one had formally checked whether calipers were actually as accurate as digital microscopy.
The study enrolled patients undergoing elective colonoscopy at a major university hospital in Montreal, Canada. A total of 143 polyps from 92 patients were included. The average patient age was 66, and about 42 percent were female.
After a polyp was removed using a technique called cold snare polypectomy, the fresh specimen was immediately measured by one of five trained research assistants using a vernier digital caliper. The caliper was recalibrated before each new measurement.
The same specimens were then photographed under a digital microscope (Dino-Lite), which provided magnification of 10x to 140x. At a later date, a separate researcher who did not know the caliper results measured each polyp image using the microscope's software. This blinded approach ensured the two measurements could be compared fairly.
Larger polyps removed by techniques other than cold snare, fragmented specimens, and polyps with unclear borders were excluded, which is why the study focused specifically on small and diminutive polyps (generally 10 mm or less).
Compared to the digital microscopy reference standard, caliper measurements showed a mean difference of only -0.22 mm. This means calipers very slightly underestimated size on average, but the difference was tiny and fell well within the clinically acceptable margin of 0.5 mm.
The study also measured how well the two methods agreed overall, using a statistical tool called the intraclass correlation coefficient (ICC). The ICC between calipers and microscopy was 0.88, which is considered good agreement. By comparison, doctors visually estimating polyp size during the procedure showed only moderate agreement (ICC around 0.52 to 0.55) with both methods.
When polyps were sorted into clinical size categories (diminutive, 5 mm or less, versus small, greater than 5 mm), calipers placed polyps in the correct category 94.4 percent of the time. This high level of categorical accuracy means caliper measurements would rarely lead to a wrong clinical decision about surveillance intervals.
After a colonoscopy, the doctor uses polyp size to decide how soon the patient should return for a follow-up scope. Polyps larger than 10 mm require closer monitoring. Getting these size categories right matters enormously for patients, as it determines how intensively their colorectal cancer risk is managed.
This study confirms that a simple, low-cost digital caliper, used in the endoscopy suite right after a polyp is removed, gives results accurate enough for these clinical decisions. Patients and their families can have confidence that size-based decisions are on firm scientific footing.
The validation also has practical implications for post-polypectomy surveillance guidelines. Guidelines set thresholds at 5 mm and 10 mm for determining how soon a follow-up colonoscopy is needed. Because calipers correctly categorized polyps 94.4 percent of the time, they are reliable tools for applying these thresholds.
One of the most exciting applications of this finding is in artificial intelligence. AI-based tools are being developed to automatically estimate polyp size during colonoscopy, which could reduce the well-known inaccuracy of visual estimation by endoscopists.
For these AI algorithms to be reliable, they must be trained and tested on accurate data. This study validates that caliper-based measurements can serve as trustworthy ground-truth data for building and evaluating AI polyp sizing systems. Previous AI studies that used calipers as their reference can now be viewed with greater confidence.
Researchers have noted that training AI on inaccurate measurements, whether from visual estimates or from post-fixation changes in pathology specimens, can bake systematic errors into the algorithm permanently. Caliper measurements, now formally validated, offer a practical and scalable alternative for generating large, high-quality annotated datasets needed for AI development.
The study was conducted at a single hospital, which may limit how broadly the results apply to other centers with different equipment or staff. The researchers used multiple research associates to perform caliper measurements, which introduces some possibility for variation between measurers, though the overall accuracy remained high.
Because only polyps removed by cold snare en bloc were included, larger polyps removed by more complex methods were not studied. This means the validation is specifically confirmed for small and diminutive polyps, the very ones that AI sizing systems are most often asked to measure.
The authors also note that digital microscopy remains the preferred reference for high-stakes multi-center validation efforts because it creates a permanent photographic record that can be audited. Calipers lack this documentation feature, which is a real limitation in research settings where independent verification is important.
Future work should evaluate caliper accuracy in larger polyps and across multiple centers to extend the validation to a broader range of clinical scenarios.
This study provides the first formal scientific validation that digital caliper-based measurement of small and diminutive colorectal polyps is accurate, reproducible, and comparable to the established digital microscopy reference standard.
The findings mean that caliper measurement, already widely used in clinical practice and research, can now be used with confidence as a reference standard for new endoscopic tools and AI algorithms. This helps ensure the ongoing science of colorectal cancer prevention rests on a solid methodological foundation.
For patients, this means that the measurements doctors and researchers rely upon for colonoscopy follow-up decisions and AI tool development are scientifically sound. Accurate measurement is a small but important link in the chain of detecting and preventing colorectal cancer before it becomes life-threatening.