Introduction

Computer-aided diagnosis (CADx) allows the real-time prediction of polyp histology during colonoscopy based on their appearance. This systematic review aims to compare the diagnostic performance of two future CADx utilization: the autonomous CADx strategy (CADx-alone) and the endoscopist with CADx assistance strategy (CADx-assisted).

Aims & Methods

We searched MEDLINE, EMBASE, and Scopus databases published before 20 March 2024. We screened for histologically-verified accuracy diagnostic studies that compared the real-time optical diagnosis performance of CADx-alone strategy versus CADx-assisted strategy for polyps ≤5mm. Only high confidence optical diagnosis were condsidered for the CADx-assisted optical diagnosis. A random effect model was used to calculated the pooled sensitivity, specificity, negative predictive value, positive predictive value and accuracy. Two separate analyses were performed: one on polyps in the whole colon and another restricted to rectosigmoid polyps.

Results

The analysis was based on seven studies. The entire colon, included 5,765 polyps for CADx-alone and 5,009 polyps for CADx-assisted; CADx-alone showed a significant reduction in sensitivity (88% vs. 93%; RR: 0.95; 95% CI: 0.92, 0.99; I²: 79.46%), negative predictive value (NPV) (85% vs. 90%; RR: 0.94; 95% CI: 0.91, 0.98; I²: 74.96%), PPV (81% vs. 83%; RR: 0.98; 95% CI: 0.96, 1.00; I²: 0%) and overall accuracy (84% vs. 86%; RR: 0.96; 95% CI: 0.95, 0.98; I²: 0%). However, there was no significant difference in specificity (74% vs. 75%; RR: 0.98; 95% CI: 0.94, 1.02; I²: 55.78%). When the analysis was restricted to rectosigmoid polyps, which included 2,515 polyps evaluated by CADx-alone and 2,153 by CADx-assisted, CADx-alone showed significantly lower specificity (83% vs. 85%; RR: 0.97; 95% CI: 0.95, 1.00; I²: 26.36%), PPV (73% vs. 79%; RR: 0.94; 95% CI: 0.89, 1.00; I²: 15.50%) and overall accuracy (86% vs. 88%; RR: 0.97; 95% CI: 0.95, 0.98; I²: 0%). No significant difference was observed in sensitivity (87% vs. 92%; RR: 0.96; 95% CI: 0.92, 1.01; I²: 33.60%) and NPV (94% vs. 95%; RR: 0.99; 95% CI: 0.97, 1.00; I²: 0%)

Conclusion

Our findings show that autonomous CADx failed to reach the CADx-assisted sensitivity, NPV, PPV and accuracy in the whole colon. When restricted to rectosigmoid polyps, autonomous CADx failed to reach specificity, PPV and accuracy. These findings underscore the necessity for further enhancements in the autonomous CADx diagnostic performance and the need for further studies that investigate the human-machine interaction.

Disclosure

Cesare Hassan: Fujifilm Co. (consultancy); Medtronic Co. (consultancy), Yuichi Mori: Olympus Corp (consultancy, speaking honorarium, equipment loan); Cybernet System (ownership interest), Alessandro Repici: Fujifilm Co. (consultancy); Olympus Corp (consultancy); Medtronic Co. (consultancy).

Introduction

The Resect and Discard strategy allows endoscopists to replace post-polypectomy pathology with real-time prediction of polyp histology during colonoscopy (optical diagnosis). The present systematic review and meta-analysis investigated how implementation of computer-aided diagnosis (CADx) for polyp pathology affected this optical diagnosis procedure.

Aims & Methods

We conducted a search of MEDLINE, EMBASE, and Scopus databases, covering studies published from inception to February 20, 2024. We included histologically verified accuracy diagnostic studies that evaluated the real-time optical diagnosis performance of CADx systems for polyps ≤5mm in the entire colon. The aim was to estimate the potential benefits and harms associated with integrating CADx assistance into the endoscopist optical diagnosis, both before and after its implementation. The primary outcomes were the proportion of polyps that would have avoided pathological assessment (main benefit) and the proportion of polyps incorrectly predicted (main harm) due to false-positive or -negative optical diagnosis. Only diagnosis with a high-level of confidence were considered.

Results

A total of 11 studies with 7,400 polyps were eligible for analysis, providing stand-alone performance of CADx with sensitivity and specificity in identifying adenomatous histology of 87% (95% confidence interval [CI]: 0.82, 0.91) and 75% (95% CI: 0.68, 0.82), respectively. Three studies (4,086 polyps) were included to analyze the primary benefit/harm outcome measures. There was no significant difference between the CADx-unassisted and CADx-assisted strategies for the proportion of polyps that would have avoided pathological assessment (90% vs. 90%; risk ratio: 1.01; 95% CI: 0.99, 1.04; I²=53.49%; moderate certainty evidence). CADx-assisted strategy reduced the proportion of incorrectly predicted polyps as compared with CADx-unassisted (13% vs. 12%; risk ratio: 0.88; 95% CI: 0.79, 0.98; I²=0%; moderate certainty evidence) mainly due to a decrease in false-positive diagnosis.

Conclusion

CADx implementation did not result in clinically relevant benefits for the Resect-and-Discard strategy.

Disclosure

Cesare Hassan: Fujifilm Co. (consultancy); Medtronic Co. (consultancy), Yuichi Mori: Olympus Corp (consultancy, speaking honorarium, equipment loan); Cybernet System (ownership interest), Alessandro Repici: Fujifilm Co. (consultancy); Olympus Corp (consultancy); Medtronic Co. (consultancy).

Introduction

AI (artificial intelligence)-driven computer-aided detection (CADe) systems are typically trained using high-quality, standardized colonoscopy images/videos, assuming its use under optimal bowel preparation. However, it remains uncertain how varying levels of bowel cleanliness affect CADe performance.

Aims & Methods

To measure the impact of bowel preparation quality on the adenoma detection capability of CADe systems.
This is a cumulative post-hoc analysis of individual data from three randomized controlled studies with similar design comparing CADe-assisted with standard colonoscopy (SC). Only patients with adequate bowel preparation (Boston Bowel Preparation Scale [BBPS] >2 in each colonic segment) were included. The primary outcome measure was the difference in adenoma detection rate (ADR) and adenoma per colonoscopy (APC) between the CADe-assisted and SC groups, according to the quality level of bowel preparation, classified as good (BBPS 6–7) or excellent (BBPS 8–9).

Results

A total of 2103 patients (1047 CADe, 1056 SC; mean age 61.5 ± 9.2 years; 50.4% male) were included. Bowel preparation was rated as good in 906 patients (43.1%) and excellent in 1197 patients (56.9%). CADe increased ADR and APC compared to SC in both good (52.2% vs. 43.8%; 1.21 vs. 0.93) and excellent (54.5% vs. 43.9%; 1.09 vs. 0.89) bowel preparation groups. The absolute increase in ADR with CADe was +8.4% (95% CI: 1.9–14.9) and +10.6% (95% CI: 5.0–16.2) in good and excellent preparation groups, respectively, with no significant interaction between preparation quality and CADe effect (Δ = +2.2%, 95% CI: -6.4–10.8; interaction p = 0.612). Similarly, APC increases were +0.28 (95% CI: 0.06–0.50) and +0.20 (95% CI: 0.03–0.36) for the good and excellent preparation groups, respectively, without significant interaction observed (Δ = -0.08, 95% CI: –0.22 to +0.06, interaction p = 0.261). Similar trends were observed when the analysis was restricted to the right colon or to diminutive adenomas.

Conclusion

In patients with adequate bowel preparation, CADe improves adenoma detection regardless of cleansing quality grades.

References

1. Repici, A, Badalament M, Maselli R et al. Efficacy of Real-Time Computer-Aided Detection of Colorectal Neoplasia in a Randomized Trial. Gastroenterology 2020; 159: 512–520
2. Repici A, Spadaccini M, Antonelli G, et al. Artificial intelligence and colonoscopy experience: lessons from two randomised trials. Gut 2022; 71: 757-765.
3. Rondonotti E, Di Paolo D, Rizzotto ER, et al. Efficacy of a computer-aided detection system in a fecal immunochemical test-based organized colorectal cancer screening program: a randomized controlled trial (AIFIT study). Endoscopy. 2022; 54: 1171-1179.

Introduction

Guidelines recommend independent auditing of diagnostic performance for clinical implementation of computer-aided optical polyp diagnosis (CADx).

Aims & Methods

This study aimed to evaluate diagnostic performance and interobserver agreement of auditors, and to offer guidance on conducting CADx audits. Images and videos of all ≤5-mm polyps from a large prospective study with systematic activation of CADx were audited by three expert endoscopists. Experts performed independent, blinded diagnostic review including documentation of confidence level. The primary outcome was sensitivity of audit by three experts for high-confidence adenomas compared with pathology. Secondary outcomes included the number of reviewers for optimal CADx auditing and interobserver agreement.

Results

487 diminutive polyps were audited (510 patients). Sensitivity was 99.4% (95%CI 96.0–100) using three experts (Strategy A); 88.7% (95%CI 84.1–92.1) using two experts and one referee (Strategy B); 99% (95%CI 96–99.8), 98.8% (95%CI 95.4–99.8), and 99.4% (95%CI 96.3–100) using two-expert combinations (Strategy C); and 98.2% (95% CI 95.1-99.4), 97.3% (95% CI 94.0-98.9), and 88.9% (95% CI 83.6-92.7) for each expert individually (Strategy D). Among 266 pathology-based adenomas, Strategy A evaluated 160 polyps versus 196, 172, and 170 in Strategy C; and 220, 223, and 207 in Strategy D. Strategy B evaluated all 266 adenomas. Overall interobserver agreement was moderate (kappa 0.52), but very high for high-confidence adenomas (kappa 0.89).

Conclusion

Expert audit for evaluating CADx resulted in high sensitivity and interobserver agreement for high-confidence adenomas. Audit by two experts, with a third expert for arbitration, permitted audit of all polyps and could effectively assess CADx within clinical studies.

Disclosure

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

Introduction

Histopathological assessment remains the gold standard for evaluating diminutive colorectal polyps, despite the potential of optical diagnosis (OD) to reduce colonoscopy-associated costs (1). However, implementing OD strategies remains limited in clinical practice (2). This study aimed to evaluate patient acceptance of truly replacing pathology evaluation of diminutive polyps with resect and discard (RD) and diagnose and leave (DL) strategies using computer-aided polyp diagnosis (CADx).

Aims & Methods

In this prospective clinical pilot study at the Montreal University Hospital Center, 102 patients undergoing colonoscopy were invited to participate in the study. Patients received information on CADx, RD, DL and were informed that if they agree to participate in the study, we will diagnose all polyps using CADx during the colonoscopy, but polyps will be discarded or left in place and no pathology exams will be performed. For consenting patients, blue light imaging (BLI)-assisted CADx was used and endoscopists made the final histology determination and rated their confidence level (high vs. low) in the OD. Endoscopists also determined if high-risk features were present before using RD or DL strategies. Then all diminutive polyps optically diagnosed with high confidence as adenomas were resected and discarded and all diminutive rectosigmoid polyps optically diagnosed as hyperplastic with high confidence were left in situ. All polyps >5mm, with disagreement between endoscopists and CADx OD, low-confidence OD, or high-risk features (JNET 2b or higher) were sent for pathology evaluation. All cases were video recorded, and a quality review of the diagnoses performed with two expert endoscopists (DR and HP).

Results

Among the 102 patients approached, 97 (95%) accepted participating and utilization of CADx instead of pathology. Three stated that they did not want to participate in any research, and two patients stated that they did not trust OD or CADx as the reason for refusal. Overall, 164 diminutive polyps were removed in 97 patients (mean age=67.2 years), of which 36% were in the rectosigmoid and 64% in the proximal colon. CADx-assisted OD was performed on 97% of the polyps. The proportion of polyps with a low-confidence OD or disagreement between endoscopists' and CADx-based OD was 19/159 (11.9%). Overall, 98/159 (61.6%) polyps underwent the RD strategy, of which 4 polyps were discarded by the endoscopist's discretion despite disagreement between endoscopists and CADx OD. One of these 4 polyps had to undergo the DL strategy but was RD because of serrated features. One polyp in the proximal colon was resected and retrieved because of serrated features. 40 polyps were DL in the rectosigmoid, of which 1 polyp was neoplastic based on CADx and endoscopist's OD.

Conclusion

We observed a very high patient acceptance for truly replacing diminutive polyp histopathology with CADx-assisted OD. CADx-based OD was used to RD 98/159 (61.6%) or DL 40/159 (25.2%) diminutive polyps found in the cohort.

References

1) Rex, Douglas K. It’s not lack of evidence holding back resect and discard. Gastrointestinal Endoscopy, Volume 96, Issue 6, 1033 - 1035.
2) Willems P, Djinbachian R, Ditisheim S, Orkut S, Pohl H, Barkun A, Bouin M, Faulques B, von Renteln D. Uptake and barriers for implementation of the resect and discard strategy: an international survey. Endosc Int Open. 2020 May;8(5):E684-E692.

Disclosure

Dr Rex is a consultant for Olympus Corporation, Boston Scientific, Aries Pharmaceutical, Braintree Laboratories, Lumendi Ltd., Norgine, Endokey, GI Supply,
Medtronic, and Acacia Pharmaceuticals; the recipient of research support from EndoAid, Olympus Corporation, Medivators, Erbe USA Inc, and Braintree Laboratories; and a shareholder in Satisfai Health. Dr Pohl has received research funding from Steris, Cosmo, and is an advisory board member of InterVenn. Dr 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. The remaining authors disclose no conflicts.