Polyp size is crucial for determining colonoscopy surveillance intervals. We present an artificial intelligence (AI) model for colorectal polyp size measurement without a reference object.

The regression model for polyp size estimation was developed using outputs from two SegFormer models, segmentation and depth estimation. Initially built on colonoscopic images of polyp phantoms, the model underwent transfer learning with 1,304 real-world images. Testing was conducted on 178 images from 52 polyps, independent of the training set, using a snare as the ground truth for size comparison with the AI-based model. Polyps were classified into three size groups: ≤ 5 mm, 5-10 mm, and ≥ 10 mm. Error rates were calculated to evaluate discrepancies in actual size values between the AI model and the snare method. Precision indicated the positive predictive value per size group and recall and Bland-Altman were also conducted.

The Bland-Altman analysis showed a mean bias of -0.03 mm between methods, with limits of agreement from -1.654 mm to 1.596 mm. AI model error rates for actual size discrepancies were 10.74%, 12.36%, and 9.89% for the ≤ 5 mm, 5-10 mm, and ≥ 10 mm groups, respectively, averaging 11.47%. Precision values were 0.870, 0.911, and 0.857, with overall recall of 0.846.

Our study shows that colorectal polyp size measurement by AI model is practical and clinically useful, exhibiting low error rates and high precision. AI shows promise as an accurate tool for measurement without the need for a reference object during screening colonoscopy.

Accurate size measurement of colorectal polyps is critical for clinical decision making and patient management. This systematic review aimed to evaluate the current techniques used for colonic polyp measurement to improve the reliability of size estimations in routine practice.A comprehensive literature search was conducted across PubMed, EMBASE, and MEDLINE to identify studies relevant to size measurement techniques published between 1980 and March 2024. The primary outcome was the accuracy of polyp sizing techniques used during colonoscopy.61 studies were included with 34 focusing on unassisted and assisted endoscopic visual estimation and 27 on computer-based tools. There was significant variability in visual size estimation among endoscopists. The most accurate techniques identified were computer-based systems, such as virtual scale endoscopes (VSE) and artificial intelligence (AI)-based systems. The least accurate techniques were visual or snare-based polyp size estimation. VSE assists endoscopists by providing an adaptive scale for real-time, direct, in vivo polyp measurements, while AI systems offer size measurements independent of the endoscopist's subjective judgment.This review highlights the need for standardized, accurate, and accessible techniques to optimize sizing accuracy during endoscopic procedures. There is no consensus on a gold standard for measuring polyps during colonoscopy. While biopsy forceps, snare, and graduated devices can improve the accuracy of visual size estimation, their clinical implementation is limited by practical, time, and cost challenges. Computer-based techniques will likely offer improved accuracy of polyp sizing in the near future.

Accurate assessment of colorectal polyp size is crucial for determining treatment and surveillance policies. However, visual estimation of lesion diameter is often inaccurate, making simple and effective educational tools essential. We aimed to evaluate the learning effects of virtual scale endoscopy (VSE).

Thirty-three endoscopists first watched prelearning videos for SET1. They then estimated the diameters of 20 lesions and referred to instructional videos with VSE for self-study. Subsequently, they watched the postlearning videos for SET2 and estimated the lesion diameters. The error between the estimated and correct lesion sizes of both sets was compared. To evaluate longer-term learning effects, participants answered SET3 and SET4, which consisted of the same questions as SET2 and SET1, respectively, but 2 to 3 months later without watching the instructional video for SET2.

The error in the participants' estimation of the correct lesion diameter improved from SET1 to SET2 (34.7 ± 6.6 mm vs 30.7 ± 7.7 mm, P = .048), with a significant learning effect and error improvement specifically among nonexperts (35.2 ± 5.3 mm vs 30 ± 6.8 mm, P = .028). In SET3 and SET4, participants' errors indicated that the learning effect was well maintained (SET2 vs SET3: 30.7 ± 7.7 mm vs 28.6 ± 7.2 mm [P = .1]; SET1 vs SET4: 34.7 ± 6.6 mm vs 31.7 ± 7.1 mm [P = .025]).

VSE videos are a valuable learning tool for estimating lesion diameter, particularly for novice endoscopists, both in the short and longer term.

Measurement of colorectal polyps is typically performed through visual estimation, which is prone to bias. Studies have evaluated the accuracy of visual estimation and utility of assistive tools, but results have been mixed. The aim of this study was to clarify the accuracy of visual estimation as a measurement tool and the benefits of artificial intelligence.

MEDLINE and Embase were searched through October 2024. Extraction and quality assessment were performed independently by 2 authors. The primary outcome was the pooled absolute mean difference in size between visual estimation and control. Secondary outcomes included subgroup analysis of expert vs trainee status, accuracy of artificial intelligence, study origin (East vs West), comparator type, definition of accuracy, polyp size, direction of estimation, and image type.

Thirty-five studies with 42,964 polyp measurements were included in our analysis. All studies were of high quality, and there was no evidence of publication bias. The pooled absolute mean difference from comparator was 1.68 mm (confidence interval 1.21-2.15) with high variability explained by differences in the comparator, direction of estimation, image type, and size of the polyp. Overall accuracy was 60% with high variability as well, with increased accuracy with video displayed over photographs. Artificial intelligence improved accuracy with an odds ratio of 7.46.

Visual estimation is an inaccurate and imprecise way to measure colorectal polyps. Further research is needed to determine the impact on clinical outcomes related to colorectal cancer. Investment in new technology to aid in polyp measurement is an important next step.

The virtual scale endoscope (VSE) is a newly introduced endoscope that helps endoscopists in measuring colorectal polyp size (CPS) during colonoscopy by displaying a virtual scale. This study aimed to determine the usefulness of the VSE for CPS measurement and the educational benefit of using VSE images to improve CPS estimation accuracy.

This study included 42 colorectal polyps in 26 patients treated at Hiroshima University Hospital. In study 1, CPS measured using a VSE before endoscopic mucosal resection was compared with CPS measured on resected specimens, and the agreement between the two measurement methods was evaluated via Bland-Altman analysis. In study 2, 14 endoscopists (5 beginners, 5 intermediates, and 4 experts) took a pre-test to determine the size of 42 polyps. After the pre-test, a lecture on CPS measurement using VSE images was given. One month later, the endoscopists took a post-test to compare CPS accuracy before and after the lecture.

In study 1, Bland-Altman analysis revealed no fixed or proportional errors. The mean bias ±95% limits of agreement (±1.96 standard deviations) of the measurement error was -0.05 ± 0.21 mm, indicating that the agreement between two measurement methods was sufficient. In study 2, the accuracy of CPS measurement was significantly higher among beginners (59.5% vs. 26.7%, p < 0.01) and intermediates (65.2% vs. 44.3%, p < 0.05) in the post-test than in the pre-test.

The VSE accurately measures CPS before resection, and its images are useful teaching tools for beginner and intermediate endoscopists.

An accurate polyp size estimation during colonoscopy is crucial to determine the surveillance interval and predict the risk of malignant progression. However, there is a high degree of subjectivity in estimating polyp size among endoscopists in clinical practice. We aimed to assess the efficacy of a novel method that uses artificial intelligence (AI) to measure the size of colon polyps and compare it with current approaches.

Using the W-Net model for vessel segmentation and based on retinal image datasets (DRIVE, STARE, CHASE-DB, and HRF) and colonoscopy images, we developed the bifurcation-to-bifurcation (BtoB) distance measuring method and applied it to endoscopic images. Measurements were compared with those obtained by eight endoscopists (four expert and four trainees). Diagnostic ability and reliability were evaluated using Lin's concordance correlation coefficients (CCCs) and Bland-Altman analyses.

For both experts and trainees, visually estimated sizes of the same polyp were significantly inconsistent depending on the camera view used (P < 0.001). Bland-Altman analyses showed that there was a trend toward underestimation of the sizes of the polyps in both groups, especially for polyps larger than 10 mm. The new technique was highly accurate and reliable in measuring the size of colon polyp (CCC, 0.961; confidence interval 0.926-0.979), clearly outperforming the visual estimation and open biopsy forceps methods.

The new AI measurement method improved the accuracy and reliability of polyp size measurements in colonoscopy images. Incorporating AI might be particularly important to improve the efficiency of trainees at estimating polyp size during colonoscopy.

Accurate polyp size estimation is essential in deciding the therapeutic strategy of colorectal polyps and endoscopic surveillance intervals. However, many endoscopists frequently make incorrect size estimations without being aware of their errors. This cross-sectional study aimed to clarify the characteristics of endoscopists associated with inaccurate estimation.

We previously conducted a web trial involving 261 endoscopists in 51 institutions in Japan to assess their ability to estimate polyp size. Participants answered questions about polyp size using visual estimates in a test involving images of 30 polyps. Here, we investigated the relationships between inaccurate size estimation and the backgrounds of participants. The rates of overestimation and underestimation of polyp size were also compared to clarify any trends in the answers of participants with low accuracy (< 50%).

Multivariable logistic regression analysis revealed that the number of colonoscopic procedures in the past year was the only factor associated with a low accuracy of polyp size estimation (odds ratio 0.750, 95% confidence interval 0.609-0.925; P = 0.007). Endoscopists with low accuracy had a greater tendency to overestimate polyp size (42.3% overestimation and 21.2% underestimation, P < 0.001) compared with other endoscopists (16.6% overestimation and 17.9% underestimation, P = 0.951).

Endoscopists with limited experience of colonoscopy in the past year were more likely to make frequent errors in size estimation. Furthermore, endoscopists making inaccurate size estimations had a propensity to overestimate polyp size.

Accurate polyp size estimation is necessary for appropriate management of colorectal polyps. Polyp size is often determined by subjective visual estimation in clinical situations; however, it is inaccurate, especially for beginner endoscopists. We aimed to clarify the usefulness of our short training video, available on the Internet, for accurate polyp size estimation.

We conducted a multicenter prospective controlled study in Japan. After completing a pretest composed of near and far images of 30 polyps, participants received the educational video lecture (<10 min long). The educational content included the knowledge of strategies based on polyp size and criteria for size estimation including the endoscopic equipment size and videos of polyps in vivo. After one month, the participants undertook a posttest. The primary outcome was a change in the accuracy of polyp size visual estimation between the pretest and posttest in beginners.

Participants including 111 beginners, 52 intermediates, and 97 experts from 51 institutions completed both tests. Accuracy of polyp size estimation in the beginners showed a significant increase after the video lecture [54.1% (51.3-57.0%) to 59.0% (56.5-61.5%), P = 0.003]. Multivariable logistic regression analysis showed that the category of beginners and a low score on pretest (P = 0.020 and <0.001, respectively) were the factors that contributed to an increase of ≥10% in the accuracy.

Our educational video led to an improvement in polyp size estimation in beginners. Furthermore, this video may be useful for non-beginners with insufficient polyp size estimation accuracy.

The diversity, technical skills required, and risk inherent to advanced endoscopy techniques all contribute to complex training curricula and steep learning curves. Since trainees develop endoscopy skills at different rates, there has been a shift towards competency-based training and certification. Validated endoscopy performance measures for trainees are, therefore, necessary. The aim of this systematic review was to describe and critically assess the existing evidence regarding measures of performance for trainees in advanced endoscopy.

A systematic review of the literature from January 1980 to January 2016 was carried out using the MEDLINE, EMBASE, CENTRAL, and ISI Web of knowledge databases. MeSH terms related to 'advanced endoscopy' and 'performance' were applied to a highly sensitive search strategy. The main outcomes were face, content, and construct validity, as well as reliability.

The literature search yielded 1,662 studies and 77 met the inclusion criteria after abstract and full-text review (endoscopic retrograde cholangiopancreatography (ERCP)=23, endoscopic ultrasound (EUS)=30, colonoscopic polypectomy (CP)=11, balloon-assisted enteroscopy (BAE)=7, luminal stenting=3, radiofrequency ablation (RFA)=2, and endoscopic muscosal resection (EMR)=1). Good validity and reliability were found for measurement tools of overall performance in ERCP, EUS and CP, with applications for both patient-based and simulator training models. A number of specific technical skills were also shown to be valid measures of performance. These include: selective biliary cannulation, sphincterotomy, biliary stent placement, stone extraction and procedure time for ERCP; pancreatic solid mass T-staging, EUS-guided fine needle aspiration (EUS-FNA) procedure time, number of EUS-FNA passes and puncture precision for EUS; procedure time and en bloc resection rate for CP; retrograde fluoroscopy time for BAE; and mean number of endoscopy sessions required to achieve complete eradication of intestinal metaplasia (CIEM) for RFA. The evidence for EMR and luminal stenting is of insufficient quality to make recommendations.

We have identified multiple valid and readily available performance measures for advanced endoscopy trainees for ERCP, EUS, CP, BAE and RFA procedures. These tools should be considered in advanced endoscopy training programs wishing to move away from apprenticeship-based training and towards competency-based learning with the help of patient-based and simulator tools.

To identify whether the forceps estimation is more useful than visual estimation in the measurement of colon polyp size.

We recorded colonoscopy video clips that included scenes visualizing the polyp and scenes using open biopsy forceps in association with the polyp, which were used for an exam. A total of 40 endoscopists from the Busan Ulsan Gyeongnam Intestinal Study Group Society (BIGS) participated in this study. Participants watched 40 pairs of video clips of the scenes for visual estimation and forceps estimation, and wrote down the estimated polyp size on the exam paper. When analyzing the results of the exam, we assessed inter-observer differences, diagnostic accuracy, and error range in the measurement of the polyp size.

The overall intra-class correlation coefficients (ICC) of inter-observer agreement for forceps estimation and visual estimation were 0.804 (95%CI: 0.731-0.873, P < 0.001) and 0.743 (95%CI: 0.656-0.828, P < 0.001), respectively. The ICCs of each group for forceps estimation were higher than those for visual estimation (Beginner group, 0.761 vs 0.693; Expert group, 0.887 vs 0.840, respectively). The overall diagnostic accuracy for visual estimation was 0.639 and for forceps estimation was 0.754 (P < 0.001). In the beginner group and the expert group, the diagnostic accuracy for the forceps estimation was significantly higher than that of the visual estimation (Beginner group, 0.734 vs 0.613, P < 0.001; Expert group, 0.784 vs 0.680, P < 0.001, respectively). The overall error range for visual estimation and forceps estimation were 1.48 ± 1.18 and 1.20 ± 1.10, respectively (P < 0.001). The error ranges of each group for forceps estimation were significantly smaller than those for visual estimation (Beginner group, 1.38 ± 1.08 vs 1.68 ± 1.30, P < 0.001; Expert group, 1.12 ± 1.11 vs 1.42 ± 1.11, P < 0.001, respectively).

Application of the open biopsy forceps method when measuring colon polyp size could help reduce inter-observer differences and error rates.

Accurate measurement of polyp size during colonoscopy is important because the size is a surrogate marker of cancer, but a standardized measurement technique to measure polyp size has yet to be determined. We have developed a new device "a novel calibrated hood." We assessed polyp size by visual estimation and measurement using the calibrated hood.

Patients who underwent polypectomy from November 2012 to September 2013 and who had received screening colonoscopy within 6 months prior to the polypectomy were included in this study. Polypectomy was performed attaching the calibrated hood. The endoscopist measured the polyp size using the calibrated hood. Polyp size was compared between visual estimation and measurement using the calibrated hood.

Seventy-five patients with 157 polyps were included. Seventy-seven polyps fulfilled the selection criteria. Mean polyp size by visual estimation was 6.57 ± 2.15, and by using calibrated hood was 5.94 ± 1.73 (p = 0.005). There was a significant difference between measurements using the calibrated hood vs. visual estimation by inexperienced trainees; however, there was no difference in case of well-experienced endoscopists. By visual estimation, 11 of 19 polyps were decided for ≥5 mm despite being less than 5 mm, and 5 of 58 polyps were decided for <5 mm despite being 5 mm or larger in diameter.

Visual estimation of polyp size is not accurate. It is important to measure the size by an objective way, and the calibrated hood is useful in measuring polyp size, from the standpoint of accurately determining indication for polypectomy.

Colonoscopy is the reference medical examination for early diagnosis and treatment of colonic diseases. This minimally invasive technique allows endoscopists to explore the colon cavity and remove neoplasias - abnormal growths of tissue - which may develop into malignant tumors. The size, shape and appearance of a neoplasia are essential cues for diagnostic. However, the size is difficult to estimate because the absolute scale of the observed tissue is not directly conveyed in the 2D colonoscopic images. An erroneous size estimate may lead to inappropriate treatment. There currently exist no solutions to reproducible neoplasia size measurement adapted to colonoscopy. We propose a colonoscopic size measurement system for neoplasias. By using a simple planar geometry, the key technical problem is reduced to resolving scale. Our core contribution is introducing the Infocus-Breakpoint (IB) that allows us to resolve scale from a regular colonoscopic video. We define the IB as the lower limit of the colonoscope's depth of field. The IB corresponds to a precise colonoscope to tissue distance, called the reference depth, which we calibrate preoperatively. We detect the IB intraoperatively thanks to two novel modules: deformable Blur-Estimating Tracking (BET) and Blur-Model Fitting (BMF). With our system, the endoscopist may interactively measure the length and area of a neoplasia in a 2D colonoscopic image directly. Our system needs no hardware modification to standard monocular colonoscopes, yet reaching a size measurement accuracy of the order of a millimeter, as shown on several phantom and patient datasets.

Endoscopic measurement of colorectal polyps by using open forceps is commonly used in clinical trials but is subject to several sources of error.

Assess error rates in polyp size determination with open forceps at the step of measurement after the forceps are aligned on the polyp.

This was a prospective assessment of 49 gastroenterologists who received training on 10 photographs of polyps with aligned forceps and then measured 10 additional test polyps from photographs. One of the test photographs was excluded because of incorrect forceps alignment.

Data analyzed at an academic medical center.

Photographs displayed in a webinar.

Description of rates of accurate measurements, including the fraction correct within 50% and 25% margins of error.

A total of 37% of all measurements were correct to the exact millimeter, 34% were larger, and 29% smaller compared with the reference standard. A total of 47 of 49 doctors measured all 9 polyps within 50% of the reference standard, and 21 measured all 9 correctly within a 25% error margin.

Other potential sources of error in open forceps measurement were not evaluated.

Open forceps polyp size determination is subject to error at the step of using the fully aligned forceps as a scale for measurement. A margin of error of 50% up or down is needed to prevent this step in size determination from causing errors in polyp matching in clinical trials comparing diagnosis-only imaging to colonoscopy.

Despite the frequency at which urologists endoscopically estimate lesion size, their accuracy has not been established. Our objectives were to determine the accuracy of cystoscopic and ureteroscopic estimates of lesion size using in vitro models of the urinary tract and to assess potential impacting factors.

Eleven staff urologists and 9 urology learners performed cystoscopy on a series of pig bladders containing mock papillary and flat lesions. Each provided three sets of size estimates: two using only the cystoscope to assess intraobserver agreement and the third with the aid of a ureteral catheter as a visual reference. Similar estimates were made with a flexible ureteroscope on papillary lesions within an inorganic upper urinary tract model. Differences in mean estimates and the agreement between repeated estimates were assessed.

The level of endoscopic training did not influence the mean error of estimation (MEE) for either cystoscopy or ureteroscopy regardless of lesion size and appearance. Staff and learners consistently underestimated lesion size with median errors of 34% and 43%, with excellent (median intraclass correlation coefficient [ICC] of 0.97) and fair (median ICC of 0.56) reproducibility for cystoscopy and ureteroscopy, respectively. Use of the visual reference during cystoscopy did not improve the MEE.

Urologists, regardless of their level of training, substantially underestimate lesion size by 34% to 43%. These findings are independent of lesion size and appearance, and the use of a visual reference during cystoscopy. Recognizing this tendency and adjusting estimates accordingly or improving instrumentation should improve clinical and operative decision-making.

Adenomatous polyps greater than 1 cm are defined as advanced adenomas. Inaccurate size estimation can lead to inappropriate surveillance recommendations of colorectal adenomas.

The aim of this study was to determine the impact of endoscopic polyp mis-sizing on colorectal cancer surveillance recommendations.

This is a prospective study.

This study was conducted in a gastroenterology practice at a US academic medical center.

Patients undergoing colorectal cancer screening and surveillance colonoscopies from 2010 to 2011 were included.

Endoscopic size estimates of polyps 10 to 25 mm were compared with postfixation histopathologic polyp measurements for 15 different gastroenterologists. Only adenomatous polyps removed in entirety by snare polypectomy were included in the analysis. Size variation was defined as (endoscopic estimate - histopathologic size)/(histopathologic size). Clinical mis-sizing was defined as a size variation of >33%. The mean size variation, the percentage of clinical mis-sizing, and the percentage of inappropriate surveillance recommendation due to size variation >33% were reported per endoscopist.

: Included for analysis were 4990 procedures from 15 gastroenterologists. A total of 230 polyps from 200 patients met inclusion criteria. The average age was 62.6 years (SD 10.1), and 52% were men. The mean size variation between the endoscopic polyp size estimation and the histopathologic polyp was 73.6% (range of mean size variation, 13%-127%). 62.6% (range, 0%-91%) of included polyps had clinical mis-sizing. Of included polypectomies, 35.2% (range, 0%-67%) resulted in an inappropriate surveillance recommendation due to clinical mis-sizing even after considering histology and synchronous polyps.

This was a single-center study.

There is marked variation in endoscopists' ability to accurately size adenomatous polyps. Some endoscopists rarely mis-size adenomas, and their surveillance recommendations are appropriate in regard to sizing. However, other endoscopists inaccurately size adenomas, and this leads to inappropriate surveillance of colorectal polyps. In this study, approximately 1 of 3 included polypectomies yielded inappropriate surveillance recommendations because of clinical mis-sizing.

To assess the frequency of advanced colorectal adenomas in consulting patients in Iceland.

The histological configuration of colorectal adenomas (CRA) found in 3603 patients was classified into tubular (TA), villous (VA) and serrated (SA) and the degree of neoplastic severity into low-grade dysplasia (LGD), high-grade dysplasia (HGD), carcinoma in situ (CIS), intramucosal carcinoma (IMC) and submucosal carcinoma (SMC). Advanced CRA were those showing HGD, CIS, IMC and/or SMCs. In patients with two or more adenomas, the adenoma with the highest degree of epithelial neoplasia was selected to record cases.

Between 2003 and 2006 a total of 19424 endoscopic examinations (13572 colonoscopies and 5852 sigmoidoscopies) were performed in Iceland (mean, 4856 endoscopies per year). At histology a mean of 759.3 CRA per year were found. Thus, CRA were found in 15.6% of the colorectal endoscopies performed per year. Out of the 3037 CRA studied, 67% were TA, 29% VA and the remaining 4% SA. LGD was present in 79%, HGD in 15%, CIS in 2.4%, IMC in 1.9% and SMC in 1.9%. Consequently, out of 3037 CRA investigated, 652 (21.5%) were advanced CRA; 71% of these showed HGD, 11% CIS, 9% IMC and 9% SMC. Two-thirds of the 652 advanced CRA were advanced VA, and more than three-quarters of 58 advanced CRA with SMC, were advanced VA.

Advanced VA displaying intraepithelial neoplasia (HGD and CIS) showed a propensity to evolve into invasive carcinoma. Accordingly, VA displaying HGD and CIS might be regarded as biological markers for predicting colorectal cancer risk. This is the first study in which the frequency of CRA and advanced CRA detected in consulting patients is reported on a nationwide basis.