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

High-definition anorectal manometry (HD-ARM), balloon expulsion test (BET) and magnetic resonance defecography (MR-defecography) are important tools in studying constipation.

Aims & Methods

Evaluate the prevalence of functional defecation disorders and structural anorectal alterations in patients with constipation and compare the findings of HD-ARM and BET with MR-defecography.
Retrospective cohort study including patients with functional constipation (FC) or irritable bowel syndrome with constipation (IBS-C) submitted to HD-ARM, BET and MR-defecography. Patients were divided into three groups: 1) normal anorectal evacuation pattern in HD-ARM and BET; 2) abnormal anorectal evacuation pattern and BET; 3) normal anorectal evacuation pattern and abnormal BET, or abnormal anorectal evacuation pattern and normal BET.

Results

A total of 64 patients were included, 49 with FC (76.6%) and 15 with IBS-C (23.4%). There were 20 patients in group 1 (31.3%), 11 in group 2 (17.2%) and 33 in group 3 (51.5%). In MR-defecography, 47 patients had relevant structural anorectal alterations (73.4%) and 15 had an abnormal contraction of the puborectalis during the defecatory maneuver (23.4%).
Nineteen patients were diagnosed with dyssynergic defecation (29.7%). These patients had significantly higher residual anal pressure and lower percentage of anal relaxation during defecation (109.6±43.2 vs 85.8±29.7 mmHg, p=0.038 and 9% vs 24%, p=0.018, respectively), as well as a lower differential of rectoanal pressure (-32.2±31.4 vs 3.0±29.6 mmHg, p<0.001).
Patients with structural anorectal alterations had a significantly higher body mass index (BMI) (27.4±4.5 vs 23.5±3.3, p=0.002), and all obese patients had structural alterations.
Comparing the findings of HD-ARM and BET with MR-defecography: in group 1, 15 patients had structural alterations (75.0%); in group 2, 8 patients had structural alterations (72.7%); in group 3, 22 patients only had structural alterations (66.7%), 6 only had dyssynergic defecation (18.2%), 2 patients had both (6.0%) and 3 did not have any findings (9.1%). In 24.2% of patients with inconclusive findings in HD-ARM and BET, MR-defecography allowed the diagnosis of dyssynergic defecation.

Conclusion

Patients with structural anorectal alterations had a significantly higher BMI. Almost three quarters of patients with FC and IBS-C had relevant structural anorectal alterations and one third had dyssynergic defecation. Therefore, it is important to combine MR-defecography with HD-ARM and BET in the study of constipation.

Introduction

The LiverRisk score has been developed as blood-based tool to estimate liver stiffness measurement (LSM) in the general population (i.e., in subjects without known chronic liver disease (CLD), and thus, detect compensated advanced chronic liver disease (cACLD; LSM ≥10kPa) and stratify their risk of liver-related events. Notably, its utility in other settings is unclear as a one-size-fits-all blood-based test. We aimed to evaluate its diagnostic/prognostic performance in patients referred to specialized hepatology outpatient clinics (i.e., tertiary care).

Aims & Methods

Subjects with known/suspected CLD at first LSM (cohort-I, n=5897) or first referral & LSM (cohort-II, n=1558) were retrospectively included. Calibration/agreement of the LiverRisk score with LSM was assessed as well as its diagnostic accuracy for LSM-defined cACLD compared to FIB-4/APRI. Finally, its predictive performance for hepatic decompensation was evaluated.

Results

In cohort-I/II, mean age was 48.3/51.8years, 44.2%/44.7% were female, predominant etiologies were viral hepatitis (51.8%)/MASLD (63.7%), median LSM was 6.9 [5.1-10.9]/ 5.8 [4.5-8.8]kPa, and 1690 (28.7%)/ 322 (20.7%) had cACLD.
Despite a moderate correlation (Pearson’s r: 0.325/0.422), the LiverRisk score systematically underestimated LSM (mean difference according to the Bland-Altman-method: -2.9/-1.8 points≙kPa). Variation of observed differences was considerable (95%CI of agreement: 39.4 and 35.2 points≙kPa), and showed a clear pattern towards larger disagreement at higher values with lower LiverRisk score than LSM.
The diagnostic accuracy of the LiverRisk for cACLD (AUROC: 0.757 [95%CI: 0.744-0.770]/ 0.790 [95%CI: 0.762-0.819]) was comparable to FIB-4 (0.769/0.813) and APRI (0.747/0.765). The cut-off of 10points≙kPa yielded an overall accuracy of 74.2%/81.2%, high specificity (91.9/93.4%), but low negative predictive value (76.6/84.5%) with false-negative results in 20.0%/13.5% in cohort-I/II, respectively.
In cohort-I, 208 (3.5%) patients developed hepatic decompensation (median follow-up: 4.7years). The LiverRisk score showed a reasonable accuracy for predicting hepatic decompensation within 2 years (AUROC: 0.817); however, it was inferior to LSM (0.891, p<0.001) and FIB-4 (0.912, p<0.001). The strata of LiverRisk score (and of LSM/FIB-4) discriminated patient groups of distinct risks for hepatic decompensation.

Conclusion

While being a major step forward for screening subjects without known liver disease in primary care, the LiverRisk score was not superior (vs. FIB-4) for diagnosis of cACLD or (vs. FIB-4/LSM) for prediction of hepatic decompensation in a tertiary care setting, i.e., in cohorts with a higher pre-test probability of cACLD. Notably, FIB-4 even outperformed the LiverRisk score in terms of prognostic utility.