Introduction

Effective withdrawal time (EWT), as measured by EndoScreener QC (Wision A.I.), is a novel artificial intelligence (AI)-derived colonoscopy quality metric, which measures the actual duration of clear images obtained during scope withdrawal from cecum to rectum. We have previously demonstrated the superiority of EWT to standard withdrawal time (SWT) on adenoma detection rate (ADR) in a retrospective study.(1) This prospective study determines the association between EWT and colonic lesion detection rates.

Aims & Methods

This was a prospective evaluation of the use of real-time determination of EWT, as measured by the EndoScreener QC, of endoscopists who were blinded to the output of the system. Patients were recruited in our hospital from Nov 2023 to Feb 2024. All examination was performed with high-definition endoscopes (EVIS-EXERA 290 video system, Olympus Optical, Tokyo, Japan) under white light. In all examination, colonoscope was first advanced to the cecum as confirmed by identification of the appendiceal orifice and ileocecal valve or by intubation of the ileum. After cecal intubation, the colonoscopy was slowly withdrawn. The AI derived (AI) real time effective mucosal examination monitor system (EndoScreen QC) was initiated during scope withdrawal, starting from cecum to anus. The SWT was defined as total withdrawal time from cecum to rectum after excluding polypectomy and biopsy time. All detected polyps were removed during scope withdrawal only. The size (measured with biopsy forceps), location and morphology of each polyp was recorded by an independent observer. The bowel preparation quality was graded according to the Boston Bowel Preparation Scale. The primary outcome was the correlation between EWT and ADR, which was determined by the Pearson correlation coefficient (r); and compared with SWT. Other outcomes included correlation between EWT and polyp detection rate (PDR) or serrated lesion detection rate (SDR).

Results

In the prospective analysis, 185 colonoscopies performed by 9 endoscopists were analyzed. The mean age was 67.4 year (SD: 10.5) and 57.3% (n=106) were male. 83.9% (n=154) were diagnostic colonoscopy. The median Boston Bowel Preparation score (BBPS) was 7. The mean ADR was 49.2% (95% CI: 41.6-56.2, n=91), mean EWT was 4.4 (95% CI: 4.1-4.6) minutes and mean SWT was 7.1 (95% CI: 6.7-7.5) minutes. The correlation between EWT and ADR was significantly higher (r=0.98, 95%CI: 0.96-1.00) than between SWT and ADR (r=0.80, 95%CI: 0.53-0.98, p<0.01). For PDR, the correlation between EWT and PDR was also significantly higher than between SWT and PDR (r=0.97, 95%CI: 0.95-1.00 vs r=0.73, 95%CI: 0.30-0.98, p=0.02). However, there was no significant difference between EWT and SWT on SDR [r=0.92, 95%CI: 0.72-1.00 vs r=0.93, 95%CI: 0.86-0.98, p=0.89].

Table 1: Baseline characteristics and outcomes of the study

Conclusion

The novel AI-derived EWT demonstrated better correlation with ADR and PDR than SWT in this prospective study. Further studies should evaluate the role of applying real-time monitoring EWT on endoscopy training and quality improvement of endoscopists with low ADR.

References

1. Lui TKL, Ko MKL, Liu JJ, Xiao X, Leung WK. Artificial intelligence-assisted real-time monitoring of effective withdrawal time during colonoscopy: a novel quality marker of colonoscopy. Gastrointest Endosc. 2024;99(3):419-27.e6.

Disclosure

The following companies had given donation to our team for research.
Olympus, Fujifilm, Medronic, Wision AI, Insign Medical Co Ltd

Introduction

Accurate determination of withdrawal time (WT) is a critical quality metric within colonoscopy. While application of artificial intelligence (AI) to standardize WT estimation has been investigated, a critical need for further prospective validation exists.

Aims & Methods

We planned a prospective examiner-blinded evaluation to compare a novel AI system’s performance to that of physician assessment in routine colonoscopy from December 2023 to March 2024. Manual frame-by-frame video annotation provided the gold standard. The AI system autonomously generated detailed image reports, which were independently assessed by four endoscopists. (ClinicalTrials.gov NCT06094270).

Results

In the 126 analyzed examinations, our AI demonstrated a lower mean absolute error for WT estimation achieved by the AI system with 2.17 minutes compared to physician assessments with 4.23 minutes, p<0.01. Achieving mean 97% satisfaction with the timeline representation and 81% overall satisfaction, the AI generated comprehensive image reports.

Conclusion

These findings establish the potential of the AI system to improve efficiency and precision in WT calculation in clinical practice. This work represents a critical step towards integrating AI-driven solutions for optimizing clinical workflows and enhancing the quality of colorectal cancer screening.

Introduction

Effective colorectal cancer prevention relies on the quality of screening colonoscopy examinations. Cecal withdrawal time is significantly associated with adenoma detection rate (ADR), which is an established quality indicator for screening colonoscopy. Therefore, we introduced a real-time cecal withdrawal timer (CWT) that alerts the physician of the elapsed time since cecal intubation.

Aims & Methods

An observational, retrospective, single-center study to evaluate CWT in screening colonoscopy tests. The CWT is activated by a vocal command when the physician achieves cecal intubation. Following activation, the elapsed time is audio-announced every minute until the end of the procedure. The ADR of colonoscopies with and without the CWT was evaluated following CWT implementation from January 2023 to February 2024. Moreover, the ADR of the entire department in this period was compared with the year before.

Results

Of 1,557 colonoscopies between January 2023 and February 2024, 908 (58.3%) were performed with the CWT with a median withdrawal time of 8.75 (IQR: 6.9-12.0) minutes. During this period, the CWT usage was greater than 50% in 11 out of 14 months. Considering only months with CWT usage over 50%, the ADR of colonoscopies with CWT was 28.3% compared to 19.2% in those without (p<0.001). Among physicians who performed at least 20 tests with and 20 tests without the CWT, 6 out of 8 demonstrated improved ADR with the cecal timer. Notably, the 2 physicians who did not improve ADR had baseline ADR exceeding 30%. The overall ADR of the department between January 2023 and February 2024 was 25.5%, compared to 26.2% in 2022 (p=0.704).

Conclusion

Initial findings of the CWT program suggest its potential in enhancing screening colonoscopy quality indicators, although long-term prospective follow-up is warranted.

Disclosure

Nothing to declare

Introduction

Withdrawal time (WT) is strongly correlated with adenoma detection rate (ADR)^1,2. Corrected withdrawal time (cWT), defined as the withdrawal time excluding periods spent on therapeutic interventions, provides a more accurate measure of mucosal inspection effort. Optimal WT targets may differ between staff and trainee endoscopists due to differences in experience and inspection technique³.

Aims & Methods

We aimed to determine the cWT required to achieve ADR benchmarks (≥35%) and compare detection metrics between these two groups.
This prospective study included 1,439 elective colonoscopies of adult patients ≥18 years, performed at Montreal University Hospital Center: 397 by eight supervised trainees and 1,042 by 12 staff endoscopists. All procedures were video recorded, and the corresponding meta-data were documented in real-time by a research assistant present in the endoscopy room. Procedures performed for the purpose of polypectomy, incomplete colonoscopies, procedures performed by endoscopists with experience of fewer than 10 colonoscopies, and procedures with technical failures in recording that prevented video analysis were excluded. Colonoscopy videos were reviewed to precisely measure cWT, defined as total withdrawal time minus therapeutic intervention durations. Performance metrics were stratified by endoscopist training level (staff vs trainees). Primary outcome was the mean cWT required to achieve ADR ≥35%, stratified by staff and trainees. Secondary outcomes were the mean per-lesion intervention time, ADR, advanced ADR (AADR), Serrated Lesion detection rate (SLDR), APC, and the incremental increase of all detection measures per minute of cWT compared between staff and trainees, adjusting for covariates including patient demographics, bowel preparation quality, indication for colonoscopy, use of computer-assisted detection, and clustering by individual endoscopist using generalized estimating equations (GEEs).

Results

Trainees demonstrated significantly longer mean cWT than staffs (10’01’’ vs 7’53’’, p<0.001). No significant difference was observed regarding mean per-polyp intervention time (1′33″ vs 1′28″, p=0.234). On average, staff endoscopists required longer mean cWT to achieve the 35% ADR benchmark compared to trainees (9'47'' [95% CI: 8'04''-12'20''] vs. 7'05'' [95% CI: 0'00''-10'05'']). Overall AADR, SSLDR, and APC were significantly higher for trainees than staff (Table 1; AADR 10.8% vs. 5.3%; SSLDR 8.1% vs 3.6%; APC 0.8 vs 0.6). Overall, the odds of adenoma detection significantly increased with each additional cWT minute for both trainees and staffs (Table 1; 6.6% overall, 6.7% for staff, and 7.0% for trainees).

Table 1. Metrics’ estimated summaries overall and by endoscopist group. Means and 95% confidence interval (CI) are raw estimates. Percent increases and corresponding p-values arise from GEEs.

Conclusion

Trainees had longer cWT and higher detection rates overall, while staff required slightly more cWT to reach ADR target of ≥35%. These findings emphasize the importance of tailored WT benchmarks based on endoscopist training level to optimize colonoscopy quality and training outcomes.

References

1. Barclay RL, Vicari JJ, Doughty AS, Johanson JF, Greenlaw RL. Colonoscopic withdrawal times and adenoma detection during screening colonoscopy. N Engl J Med. 2006;355(24):2533–2541. doi:10.1056/NEJMoa060884

2. Zhao S, Wang S, Pan P, et al. Magnitude, risk factors, and factors associated with adenoma miss rate of tandem colonoscopy: a systematic review and meta-analysis. Gastroenterology. 2019;156(6):1661–1674.e11. doi:10.1053/j.gastro.2019.01.265

3. Rex DK, Petrini JL, Baron TH, et al. Quality indicators for colonoscopy. Gastrointest Endosc. 2006;63(4 Suppl):S16–S28. doi:10.1016/j.gie.2006.02.021

Disclosure

Daniel von Renteln has received research funding from ERBE Elektromedizin GmbH, Ventage, Pendopharm, Fujifilm, and Pentax, and has received consultant or speaker fees from Boston Scientific Inc., ERBE Elektromedizin GmbH, and Pendopharm. Roupen Djinbachian has received speaker fees from Fujifilm. The remaining authors declare that they have no conflict of interest.

Introduction

Endoscopists with baseline adenoma detection rates (ADRs) ≤26% who improve their ADRs significantly reduce their patients’ risk of post-colonoscopy colorectal cancer (PCCRC)¹. The recommended withdrawal time (WT) of 8–9 minutes is associated with higher ADRs; however, it includes time spent on therapeutic interventions, which may overestimate actual inspection time and limit its accuracy as a performance metric².

Aims & Methods

We aimed to determine the corrected withdrawal time (cWT)—defined as WT minus intervention time—required for low-performing endoscopists to achieve ADR thresholds of ≥26 and ≥35.
We prospectively analyzed 1,165 the recorded videos of elective colonoscopies performed by 17 endoscopists at Montreal University Hospital Center (IRB; CER22.013/ CER25.12283). Eligible cases were consecutive adults (≥18 years) undergoing elective colonoscopy. Exclusions were complex resections, sigmoidoscopies, incomplete exams, failed recordings, and procedures by endoscopists with <10 prior cases. A research assistant was present during each procedure and documented all periprocedural metrics in real-time. Endoscopists were stratified as low (ADR ≤25%) or high performers (ADR >25%). The primary outcome was the mean cWT required for low performers to achieve an ADR ≥26%. Secondary outcomes included the cWT needed to reach ADR ≥35% and the incremental impact of each additional minute of cWT on detection metrics. Multivariable generalized estimating equation (GEE) models adjusted for patient demographics, bowel preparation, indication of colonoscopy, time of day/month, CADe use, and endoscopist effect.

Results

Low performers required a mean cWT of 10′58″ to achieve an ADR ≥26%, and 13′51″ to reach ≥35% compared to 4′01″ for high performers. Each additional minute of cWT increased ADR odds by 15.3% for low performers and 5.0% for high performers (P < .001). Similar improvements were observed across all detection metrics (Table 1). For every additional cWT minute, the odds of detecting advanced adenomas increased by 10.1%, sessile serrated lesions (SSLs) by 10.1%, and polyps by 16.5% (all P <  .001). Adenoma and polyp counts increased by 7.5% and 7.9%, respectively. Absolute gains per minute were greater in low performers, with ADR increases exceeding 3% beyond 12 minutes of cWT. After adjusting for covariates, each additional cWT minute was associated with a significant increase in the odds of detection: an increase of 6.5% (P = .002) for adenomas, 9.2% (P < .001) for advanced adenomas, 11.5% (P < .001) for SSLs, and 14.9% (P < 0.01) for polyps. With each additional cWT minute, the adenoma count increased by 7.3% and the polyp count increased by 7.8% (P < .001) (results not shown).

Table 1. Mean Detection Metrics and Their Incremental Increase Per Additional Minute of Corrected Withdrawal Time, Overall and Stratified by High vs Low Performers.

Conclusion

Longer cWT enhanced detection across all metrics, particularly for low performes. These findings support personalized cWT targets based on individual performance rather than uniform WT recommendations.

References

1. Pilonis ND, Spychalski P, Kalager M, et al. Adenoma detection rates by physicians and subsequent colorectal cancer risk. JAMA 2025;333:400–407.
2. Rex DK, Anderson JC, Butterly LF, et al. Quality indicators for colonoscopy. Gastrointest Endosc 2024;100:352–381.

Disclosure

Daniel von Renteln has received research funding from ERBE Elektromedizin GmbH, Ventage, Pendopharm, Fujifilm, and Pentax, and has received consultant or speaker fees from Boston Scientific Inc., ERBE Elektromedizin GmbH, and Pendopharm. Roupen Djinbachian has received speaker fees from Fujifilm. The remaining authors declare that they have no conflict of interest.

Introduction

In clinical practice, colonoscopy withdrawal time (CWT) is measured from the caecum to the anal verge without accounting for cleaning (washing or suctioning) and polypectomy events during withdrawal. The evidence for the correlation of CWT and adenoma detection rate (ADR) is somewhat conflicting, partially due to research relying on retrospective data, which does not account for cleaning and polypectomy events and is subject to inaccuracies in time measurement.

A randomized controlled trial recently demonstrated that ADR increases as colonoscopy inspection time (CIT) rise from 6 to 13 minutes ¹. CIT was defined as withdrawal from the caecum to the anal verge with the timer manually stopped during cleaning and polypectomy events, but CIT measurement cannot be implemented in clinical practice without dedicated staff to pause a timer during these events.

Aims & Methods

Recent years have demonstrated the ability of artificial intelligence (AI) models to measure CWT by detecting cecal landmarks and the anal verge. Our aim was to develop convolutional neural networks (CNN) to detect phases of cleaning and polypectomy during withdrawal to facilitate automated measurement of CIT.

One hundred endoscopy videos were prospectively collected for training data. We reviewed each individual frame during withdrawal (defined as frames after a cecal landmark was first visualized). Each frame during ‘cleaning’ and ‘polypectomy’ events were annotated with respective image-level labels, and the remaining withdrawal frames labelled as ‘inspection’. Cleaning events was defined as frames during suctioning of intra-colonic content or washing. Polypectomy events was defined as the first frame with visualisation of the injection needle or forceps/snare if the polyp was not lifted until completion of inspection of post-polypectomy resection margins. These annotations were referenced as ground truth.

From the 100 videos, two ResNet-101 CNNs were developed to detect cleaning (213,936 training frames) and polypectomy (56,273 training frames). The CNNs were then evaluated with 80 colonoscopy videos. This included 40 colonoscopy videos from five expert (ADR >45%) and 40 videos from six non-expert (ADR <30%) endoscopists. Consecutive procedures were used to minimize selection bias. Videos were annotated following the above methodology.

Results

For the test set of 80 procedures, the median WT documented in the endoscopy reporting system (EPIC) was 15.0 minutes (IQR 10.0–23.0), annotated ground truth median CIT was 7.1 minutes (IQR 5.3–9.6), and AI predicted median CIT was 6.9 minutes (IQR 5.1–9.0) with a correlation coefficient (r) of 98.2%. For expert procedures (n=40), these were 16.5 minutes (IQR 11.0–25.3), 7.1 minutes (IQR 5.1–8.6) and 6.9 minutes (IQR 4.9–8.8), respectively. Amongst non-expert procedures (n=40), these were 14.0 minutes (IQR 10.0–22.0), 7.0 minutes (IQR 5.8–10.7) and 7.0 minutes (IQR 5.3–10.2), respectively. The AI system correctly categorized 96.3% (77/80) of procedures as less or more than 6 minutes CIT.

Conclusion

We have demonstrated the feasibility of AI to differentiate the phases of withdrawal to facilitate automated measurement of CIT and the considerable difference in time between CIT and the CWT documented in an endoscopy reporting system.

References

1. Desai M, Rex DK, Bohm ME, et al. Impact of withdrawal time on adenoma detection rate: results from a prospective, multi-center trial. Gastrointest Endosc. 2022 Oct 10:S0016-5107(22)02039-9

Disclosure

Professor Laurence B Lovat is a minority shareholder and consultant advisor for Odin Vision. Professor Danail Stoyanov is a shareholder for Odin Vision.

Introduction

The benefit of computer-assisted detection systems (CADe) has been evaluated in several randomized controlled trials (RCTs) showing improved adenoma detection rate (ADR) and polyp detection rates (PDR) (1-3). However, the generalizability of these studies is questioned as these studies are performed in highly controlled setting and thus not representing a real-world clinical setting (4). It is also believed that the increase in ADR by CADe-assisted colonoscopy is mainly observed in lower detector endoscopists (5, 6).
We aimed to evaluate the effect of CADe-assisted colonoscopy on ADR in an actual clinical setting in a high-volume, non-tertiary endoscopy centre with high detector endoscopists.

Aims & Methods

A single-centre, observational study was conducted on all patients undergoing colonoscopy due to symptoms of colorectal cancer (CRC), screening colonoscopies for CRC, post-polypectomy surveillance, or colonoscopy due to positive results from a fecal immunochemical test in an Australian hospital between March 2023 to October 2023. Colonoscopies were performed by 19 experienced endoscopists using the validated CADe system that allows for real-time image analysis (GI Genius, Cosmo IMD Ireland). Historical procedural data regarding colonoscopies were obtained up to 10 months before the start of the study and were compared to the CADe-assisted colonoscopies in a 2:1 fashion. The primary outcome of the study was the ADR. Data were analyzed using logistic mixed effects models with random intercepts for patient and endoscopist, expressed as Odds ratio (OR) with a 95% confidence interval (CI).

Results

A total of 7798 colonoscopies were performed for 7460 patients during the timeframe, including 2813 colonoscopies using CADe and 4985 standard colonoscopies. The ADR in the CADe-assisted colonoscopy group was 52% compared to 46% in the standard colonoscopy group. There was a significant difference in ADR (OR 1.22, 95% CI 1.09 - 1.36, p<0.001) and PDR (OR 1.22, 95% CI 1.09 - 1.37, p<0.001) between CADe-assisted colonoscopy and standard colonoscopy respectively, adjusted for age, bowel preparation quality, withdrawal time and sex.

Table 1 Adenoma detection and polyp detection rate of the CADe-assisted colonoscopy group compared to the standard colonoscopy group.

Conclusion

In this real-time, observational study, CADe-assisted colonoscopy significantly improves the adenoma detection rate and polyp detection rate. Our study shows that the implementation of CADe-assisted colonoscopies in daily practice is useful, even in high-detector endoscopists.

References

1. Wang P, Liu X, Berzin TM, Glissen Brown JR, Liu P, et al. Effect of a deep-learning computer-aided detection system on adenoma detection during colonoscopy (CADe-DB trial): a double-blind randomised study. Lancet Gastroenterol Hepatol. 2020;5(4):343-51.
2. Repici A, Spadaccini M, Antonelli G, Correale L, Maselli R, et al. Artificial intelligence and colonoscopy experience: lessons from two randomised trials. Gut. 2022;71(4):757-65.
3. Hassan C, Spadaccini M, Mori Y, Foroutan F, Facciorusso A, et al. Real-Time Computer-Aided Detection of Colorectal Neoplasia During Colonoscopy : A Systematic Review and Meta-analysis. Ann Intern Med. 2023;176(9):1209-20.
4. Patel HK, Mori Y, Hassan C, Rizkala T, Radadiya DK, et al. Lack of Effectiveness of Computer Aided Detection for Colorectal Neoplasia: A Systematic Review and Meta-Analysis of Nonrandomized Studies. Clin Gastroenterol Hepatol. 2023.
5. Karsenti D, Tharsis G, Perrot B, Cattan P, Percie du Sert A, Venezia F, et al. Effect of real-time computer-aided detection of colorectal adenoma in routine colonoscopy (COLO-GENIUS): a single-centre randomised controlled trial. Lancet Gastroenterol Hepatol. 2023;8(8):726-34.
6. Mangas-Sanjuan C, de-Castro L, Cubiella J, Diez-Redondo P, Suarez A, Pellise M, et al. Role of Artificial Intelligence in Colonoscopy Detection of Advanced Neoplasias : A Randomized Trial. Ann Intern Med. 2023;176(9):1145-52.