Artificial intelligence for colorectal neoplasia detection during colonoscopy: a systematic review and meta-analysis of randomized clinical trials

BACKGROUND

The use of artificial intelligence (AI) in detecting colorectal neoplasia during colonoscopy holds the potential to enhance adenoma detection rates (ADRs) and reduce adenoma miss rates (AMRs). However, varied outcomes have been observed across studies. Thus, this study aimed to evaluate the potential advantages and disadvantages of employing AI-aided systems during colonoscopy.

METHODS

Using Medical Subject Headings (MeSH) terms and keywords, a comprehensive electronic literature search was performed of the Embase, Medline, and the Cochrane Library databases from the inception of each database until October 04, 2023, in order to identify randomized controlled trials (RCTs) comparing AI-assisted with standard colonoscopy for detecting colorectal neoplasia. Primary outcomes included AMR, ADR, and adenomas detected per colonoscopy (APC). Secondary outcomes comprised the poly missed detection rate (PMR), poly detection rate (PDR), and poly detected per colonoscopy (PPC). We utilized random-effects meta-analyses with Hartung-Knapp adjustment to consolidate results. The prediction interval (PI) and I2 statistics were utilized to quantify between-study heterogeneity. Moreover, meta-regression and subgroup analyses were performed to investigate the potential sources of heterogeneity. This systematic review and meta-analysis is registered with PROSPERO (CRD42023428658).

FINDINGS

This study encompassed 33 trials involving 27,404 patients. Those undergoing AI-aided colonoscopy experienced a significant decrease in PMR (RR, 0.475; 95% CI, 0.294-0.768; I2 = 87.49%) and AMR (RR, 0.495; 95% CI, 0.390-0.627; I2 = 48.76%). Additionally, a significant increase in PDR (RR, 1.238; 95% CI, 1.158-1.323; I2 = 81.67%) and ADR (RR, 1.242; 95% CI, 1.159-1.332; I2 = 78.87%), along with a significant increase in the rates of PPC (IRR, 1.388; 95% CI, 1.270-1.517; I2 = 91.99%) and APC (IRR, 1.390; 95% CI, 1.277-1.513; I2 = 86.24%), was observed. This resulted in 0.271 more PPCs (95% CI, 0.144-0.259; I2 = 65.61%) and 0.202 more APCs (95% CI, 0.144-0.259; I2 = 68.15%).

INTERPRETATION

AI-aided colonoscopy significantly enhanced the detection of colorectal neoplasia detection, likely by reducing the miss rate. However, future studies should focus on evaluating the cost-effectiveness and long-term benefits of AI-aided colonoscopy in reducing cancer incidence.

FUNDING

This work was supported by the Heilongjiang Provincial Natural Science Foundation of China (LH2023H096), the Postdoctoral research project in Heilongjiang Province (LBH-Z22210), the National Natural Science Foundation of China's General Program (82072640) and the Outstanding Youth Project of Heilongjiang Natural Science Foundation (YQ2021H023).

Risk Factors Related to Polyp Miss Rate of Short-Term Repeated Colonoscopy

BACKGROUND

Colonoscopy is regarded as the gold standard for colorectal cancer screening and surveillance. However, previous studies have reported large numbers of polyps were missed during routine colonoscopy.

AIMS

To evaluate polyp miss rate in short-term repeated colonoscopy and explore the related risk factors.

METHODS

A total of 3695 patients and 12,412 polyps were included in our studies. We calculated the miss rate for polyps of different sizes, pathologies, morphologies and locations, and patients of different characteristics. Univariate and multivariate logistic regression analyses were performed to evaluate risk factors related to miss rate.

RESULTS

The polyp miss rate was 26.3% and the adenoma miss rate was 22.4% in our study. The advanced adenoma miss rate was 11.0% and the proportion of missed advanced adenomas in missed adenomas sized > 5 mm was up to 22.8%. Polyps sized < 5 mm had a significantly higher miss rate. The miss rate of pedunculated polyps was lower than that of flat or sessile polyps. Polyps in the right colon were prone to be missed than that in the left colon. For older men, current smokers, individuals with multiple polyps detected in the first colonoscopy, the risk of missing polyps was significantly higher.

CONCLUSION

Nearly a quarter of polyps were missed during routine colonoscopy. Diminutive, flat, sessile, and right-side colon polyps were at higher risk of missing. The risk of missing polyps was higher in older men, current smokers, and individuals with multiple polyps detected in the first colonoscopy than their counterparts.

Clinical Evaluation of Computer-Aided Colorectal Neoplasia Detection Using a Novel Endoscopic Artificial Intelligence: A Single-Center Randomized Controlled Trial

INTRODUCTION

Computer-aided diagnostic systems are emerging in the field of gastrointestinal endoscopy. In this study, we assessed the clinical performance of the computer-aided detection (CADe) of colonic adenomas using a new endoscopic artificial intelligence system.

METHODS

This was a single-center prospective randomized study including 415 participants allocated into the CADe group (n = 207) and control group (n = 208). All endoscopic examinations were performed by experienced endoscopists. The performance of the CADe was assessed based on the adenoma detection rate (ADR). Additionally, we compared the adenoma miss rate for the rectosigmoid colon (AMRrs) between the groups.

RESULTS

The basic demographic and procedural characteristics of the CADe and control groups were as follows: mean age, 54.9 and 55.9 years; male sex, 73.9% and 69.7% of participants; and mean withdrawal time, 411.8 and 399.0 s, respectively. The ADR was 59.4% in the CADe group and 47.6% in the control group (p = 0.018). The AMRrs was 11.9% in the CADe group and 26.0% in the control group (p = 0.037).

CONCLUSION

The colonoscopy with the CADe system yielded an 11.8% higher ADR than that performed by experienced endoscopists alone. Moreover, there was no need to extend the examination time or request the assistance of additional medical staff to achieve this improved effectiveness. We believe that the novel CADe system can lead to considerable advances in colorectal cancer diagnosis.

Comparing right colon adenoma detection rate during water exchange and air insufflation: a double-blind randomized controlled trial

BACKGROUND

Randomized controlled trials (RCTs) have reported that water exchange (WE) produced the highest adenoma detection rate (ADR) but did not evaluate right colon adenoma detection rate (rADR) as a primary outcome and only one of the trials employed blinded colonoscopists. The aim of our study was to determine whether, compared with air insufflation, WE significantly increases rADR and right colon serrated lesion detection rate (rSLDR) and decreases adenoma miss rate (rAMR).

METHODS

This prospective, double-blind RCT was conducted at a regional hospital in Taiwan between December 2015 and February 2020. Standard WE and air insufflation were performed. After cecal intubation, the second blinded endoscopist examined the right colon and obtained rADR (primary outcome) and rSLDR. Then, the primary colonoscopist reinserted the scope to the cecum with WE in both groups and performed a tandem examination of the right colon to obtain rAMR.

RESULTS

There were 284 patients (50.9% male, mean age 58.9 ± 9.4 years) who were randomized to WE (n = 144) or air insufflation (n = 140). The baseline characteristics were similar. The rADR (34.7% vs. 22.3%, p = 0.025), Boston Bowel Preparation Scale scores (mean, 2.6 ± 0.6vs. 2.2 ± 0.6, p < 0.001), rSLDR (18.1% vs. 7.1%, p = 0.007), and rAMR (31.5% vs. 45.2%, p = 0.038) were significantly different between WE and air insufflation.

CONCLUSIONS

The current study demonstrated a significantly higher rADR and rSLDR with the WE method performed by blinded colonoscopists. The impact of the significant findings in this report on the occurrence of interval cancers deserves to be studied.

Reducing adenoma miss rate of colonoscopy assisted by artificial intelligence: a multicenter randomized controlled trial

BACKGROUND

We have developed the computer-aided detection (CADe) system using an original deep learning algorithm based on a convolutional neural network for assisting endoscopists in detecting colorectal lesions during colonoscopy. The aim of this study was to clarify whether adenoma miss rate (AMR) could be reduced with CADe assistance during screening and surveillance colonoscopy.

METHODS

This study was a multicenter randomized controlled trial. Patients aged 40 to 80 years who were referred for colorectal screening or surveillance at four sites in Japan were randomly assigned at a 1:1 ratio to either the "standard colonoscopy (SC)-first group" or the "CADe-first group" to undergo a back-to-back tandem procedure. Tandem colonoscopies were performed on the same day for each participant by the same endoscopist in a preassigned order. All polyps detected in each pass were histopathologically diagnosed after biopsy or resection.

RESULTS

A total of 358 patients were enrolled and 179 patients were assigned to the SC-first group or CADe-first group. The AMR of the CADe-first group was significantly lower than that of the SC-first group (13.8% vs. 36.7%, P < 0.0001). Similar results were observed for the polyp miss rate (14.2% vs. 40.6%, P < 0.0001) and sessile serrated lesion miss rate (13.0% vs. 38.5%, P = 0.03). The adenoma detection rate of CADe-assisted colonoscopy was 64.5%, which was significantly higher than that of standard colonoscopy (53.6%; P = 0.036).

CONCLUSION

Our study results first showed a reduction in the AMR when assisting with CADe based on deep learning in a multicenter randomized controlled trial.

Effect of water exchange method on adenoma miss rate of patients undergoing selective polypectomy: A randomized controlled trial

BACKGROUND

Adenomas may be missed in up to 40% of screening colonoscopies. Although the water exchange (WE) method can improve ADR, as shown in several RCTs, it remains uncertain whether it can increase the detection of missing adenomas compared with standard air-insufflated (AI) colonoscopy.

METHODS

Patients aged 18-80 years who underwent selective polypectomy were randomly allocated to the WE or AI group. The primary endpoint was the adenoma miss rate (AMR), defined as the number of patients with one or more additional adenomas during the polypectomy procedure divided by the total number of patients in each group.

RESULTS

A total of 450 patients were enrolled, with 225 in each group. The overall AMRs were 45.8% (103/225) in the WE group and 35.6% (80/225) in the AI group (p = 0.035). More patients in the WE group had at least one missed adenoma in the proximal colon (38.2% vs 24.4%, p = 0.002). The adenoma-level miss rate was also higher in the WE group than in the AI group (35.1% vs 29.0%, p = 0.036). Subgroup analysis showed that patients in the WE group had more missed adenomas located in the proximal colon or with flat shapes.

CONCLUSIONS

This study confirmed that substantial adenomas were missed in patients undergoing selective polypectomy. The WE method significantly improved the detection of missed adenomas, especially those located in the proximal colon or with flat shapes. (ClnicalTrials.gov number: NCT02880748).

A review of water exchange and artificial intelligence in improving adenoma detection

Water exchange (WE) and artificial intelligence (AI) have made critical advances during the past decade. WE significantly increases adenoma detection and AI holds the potential to help endoscopists detect more polyps and adenomas. We performed an electronic literature search on PubMed using the following keywords: water-assisted and water exchange colonoscopy, adenoma and polyp detection, artificial intelligence, deep learning, neural networks, and computer-aided colonoscopy. We reviewed relevant articles published in English from 2010 to May 2020. Additional articles were searched manually from the reference lists of the publications reviewed. We discussed recent advances in both WE and AI, including their advantages and limitations. AI may mitigate operator-dependent factors that limit the potential of WE. By increasing bowel cleanliness and improving visualization, WE may provide the platform to optimize the performance of AI for colonoscopies. The strengths of WE and AI may complement each other in spite of their weaknesses to maximize adenoma detection.

Water exchange colonoscopy decreased adenoma miss rates compared with literature data and local data with CO2 insufflation: an observational study

Background

Reports showed adenoma miss rates (AMRs) of 22.5–27% in the right colon and 23.4–33.3% in the proximal colon. Missed lesions could contribute to postcolonoscopy cancers. Water exchange (WE) with near-complete removal of infused water during insertion increased adenoma detection rate but the impact on AMR had not been reported. We hypothesized that WE could reduce AMRs. Study 1 compared the AMRs of WE with literature data. Study 2 developed local AMR data with CO₂ insufflation.

Methods

The lead author attended a research seminar in 2017 on WE colonoscopy. For performance improvement, study 1 was undertaken. When data in study 1 confirmed WE produced a considerably lower AMRs in the right and proximal colon, study 2 with CO₂ insufflation was performed.

Results

Eighty-six patients completed each study. In study 1, WE removed 89% of infused water upon arrival to the cecum. The AMRs of right colon (17.5%) and proximal colon (15.5%) were considerably lower than those in the literature. Upon completion of study 2, compared with local data of CO₂ insufflation, WE showed a significantly lower AMR in the right (17.5% vs. 33.8%, P = 0.034) and proximal (15.5% vs. 30.4%, P = 0.018) colon, respectively. The major limitation was that the investigation consisted of two consecutive observational studies, not a randomized controlled trial (RCT).

Conclusions

WE with near-complete (89%) removal of infused water during insertion significantly decreased AMRs in the right and proximal colon compared with literature data and those of CO₂ insufflation in our hands. The provocative data warrant confirmation in a RCT.

Trial registration

NCT03832322 (Retrospectively registered on February 2, 2019).

The Role of Behind Folds Visualizing Techniques and Technologies in Improving Adenoma Detection Rate

PURPOSE OF REVIEW

Colorectal cancer is one of the most common malignancies in the Western world and is thought to develop from premalignant polyps. Over the past decade, several behind folds visualizing techniques (BFTs) have become available to improve polyp detection. This systematic review and meta-analysis aims to compare BFTs with conventional colonoscopy (CC).

RECENT FINDINGS

In the past five years, 14 randomized controlled trials (RCTs) including 8384 patients comparing different BFTs with CC were published. The overall relative risks for adenoma detection rate, polyp detection rate, and adenoma miss rate comparing BFTs with CC were 1.04 (95% confidence interval [CI] 0.98-1.10; P = 0.15), 1.03 (95% CI 0.98-1.09; P = 0.28), and 0.70 (95% CI 0.46-1.05; P = 0.08), respectively. Other quality metrics for colonoscopy were not significantly different between BFT-assisted colonoscopy and CC either. This meta-analysis of RCTs published in the past five years does not show a significant benefit of BFTs on any of the important quality metrics of colonoscopy. The lack of additional effect of BFTs might be due to improved awareness of colonoscopy quality metrics and colonoscopy skills among endoscopists combined with improvements of conventional colonoscope technology.

Colonoscopy attachments for the detection of precancerous lesions during colonoscopy: A review of the literature

Although colonoscopy has been proven effective in reducing the incidence of colorectal cancer through the detection and removal of precancerous lesions, it remains an imperfect examination, as it can fail in detecting up to almost one fourth of existing adenomas. Among reasons accounting for such failures, is the inability to meticulously visualize the colonic mucosa located either proximal to haustral folds or anatomic curves, including the hepatic and splenic flexures. In order to overcome these limitations, various colonoscope attachments aiming to improve mucosal visualization have been developed. All of them - transparent cap, Endocuff, Endocuff Vision and Endorings - are simply mounted onto the distal tip of the scope. In this review article, we introduce the rationale of their development, present their mode of action and discuss in detail the effect of their implementation in the detection of lesions during colonoscopy.

Diagnostic yield of scope retroflexion in the right colon: A prospective cohort study

BACKGROUND

Techniques have been implemented to improve colonoscopy adenoma detection rate (ADR) in the right colon.

AIMS & METHODS

We prospectively examined the additional diagnostic yield of right colon examination with colonoscope retroflexion in consecutive, symptomatic and screening-surveillance patients. Right colon was examined in forward-view first and thereafter, retroflexion was performed to re-inspect it.

RESULTS

Right colon examination in retroflexion was achieved in 620 (92%) patients. Increased inserted scope length to the cecum (OR: 0.48 [95% CI: 0.27-0.84]) and elderly status (OR: 0.53 [95% CI: 0.430-0.94]) predicted retrofexion failure. Forward-view colonoscopy detected 134 polyps and 112 adenomas in 105 and 85 patients, respectively. Scope retroflexion revealed 7 missed (6 adenomas--2 advanced) polyps in 7 patients; indicating 4.96 (95% CI: 1.37-8.55) % and 5.1 (95% CI: 1.12-9.05) % per-polyp and per-adenoma miss rates, respectively. In ITT analysis, per-patient polyp and adenoma miss rates were 1.041% and 0.89%, respectively. Among screening-surveillance patients, retroflexion detected 3 missed adenomas (2 advanced) in 3 patients, resulting in changed surveillance schedule in 2 of them (5.12 per 1000 screening-surveillance patients). Early study termination was favored by low right colon ADR improvement and lacking substantial surveillance interval change.

CONCLUSION

The additional diagnostic yield of scope retroflexion in the right colon is questionable.

Intraprocedural bowel cleansing with the JetPrep cleansing system improves adenoma detection

AIM: To investigate the impact of JetPrep cleansing on adenoma detection rates.

METHODS: In this prospective, randomized, crossover trial, patients were blindly randomized to an intervention arm or a control arm. In accordance with the risk profile for the development of colorectal carcinoma, the study participants were divided into high-risk and low-risk groups. Individuals with just one criterion (age > 70 years, adenoma in medical history, and first-degree relative with colorectal cancer) were regarded as high-risk patients. Bowel preparation was performed in a standardized manner one day before the procedure. Participants in the intervention arm underwent an initial colonoscopy with standard bowel cleansing using a 250-mL syringe followed by a second colonoscopy that included irrigation by the use of the JetPrep cleansing system. The reverse sequence was used in the control arm. The study participants were divided into a high-risk group and a low-risk group according to their respective risk profiles for the development of colorectal carcinoma.

RESULTS: A total of 64 patients (34 men and 30 women) were included in the study; 22 were included in the high-risk group. After randomization, 30 patients were assigned to the control group (group A) and 34 to the intervention group (group B). The average Boston Bowel Preparation Scale score was 5.15 ± 2.04. The withdrawal time needed for the first step was significantly longer in group A using the JetPrep system (9.41 ± 3.34 min) compared to group B (7.5 ± 1.92 min). A total of 163 polyps were discovered in 64 study participants who underwent both investigation steps. In group A, 49.4% of the polyps were detected during the step of standard bowel cleansing while the miss rate constituted 50.7%. Group B underwent cleansing with the JetPrep system during the first examination step, and as many as 73.9% of polyps were identified during this step. Thus, the miss rate in group B was a mere 26.1% (P < 0.001). When considering only the right side of the colon, the miss rate in group A during the first examination was 60.6%, in contrast to a miss rate of 26.4% in group B (P < 0.001).

CONCLUSION: JetPrep is recommended for use during colonoscopy because a better prepared bowel enables a better adenoma detection, particularly in the proximal colon.

High definition plus colonoscopy combined with i-scan tone enhancement vs. high definition colonoscopy for colorectal neoplasia: A randomized trial

BACKGROUND

High definition endoscopy is the accepted standard in colonoscopy. However, an important problem is missed polyps.

AIMS

Our objective was to assess the additional adenoma detection rate between high definition colonoscopy with tone enhancement (digital chromoendoscopy) vs. white light high definition colonoscopy.

METHODS

In this prospective randomized trial patients were included to undergo a tandem colonoscopy. The first exam was a white light colonoscopy with removal of all visualized polyps. The second examination was randomly assigned in a 1:1 ratio as either again white light colonoscopy (Group A) or colonoscopy with tone enhancement (Group B). Primary endpoint was the adenoma detection rate during the second withdrawal (sample size calculation - 40 per group).

RESULTS

67 lesions (Group A: n=34 vs. Group B: n=33) in 80 patients (mean age 61 years, male 64%) were identified on the first colonoscopy. The second colonoscopy detected 78 additional lesions: n=60 with tone enhancement vs. n=18 with white light endoscopy (p<0.001). Tone enhancement found more additional adenomas (A n=20 vs. B n=6, p=0.006) and identified significantly more missed adenomas per subject (0.5 vs. 0.15, p=0.006).

CONCLUSIONS

High definition plus colonoscopy with tone enhancement detected more adenomas missed by white light colonoscopy.

The Effect of the Bowel Preparation Status on the Risk of Missing Polyp and Adenoma during Screening Colonoscopy: A Tandem Colonoscopic Study

Background/Aims

Although a small amount of fecal material can obscure significant colorectal lesions, it has not been well documented whether bowel preparation status affects the missing risk of colorectal polyps and adenomas during a colonoscopy.

Methods

We prospectively enrolled patients with one to nine colorectal polyps and at least one adenoma of >5 mm in size at the screening colonoscopy. Tandem colonoscopy with polypectomy was carried out within 3 months.

Results

A total of 277 patients with 942 polyps and 714 adenomas completed index and tandem examinations. At the index colonoscopy, 187 polyps (19.9%) and 127 adenomas (17.8%) were missed. The per-patient miss rate of polyps and adenomas increased significantly as the bowel cleansing rate declined from excellent to poor/inadequate on the Aronchick scale (polyps, p=0.024; adenomas, p=0.040). The patients with poor/inadequate bowel preparation were independently associated with an increased risk of having missed polyps (odds ratio [OR], 3.21; 95% confidence interval [CI], 1.13 to 9.15) or missed adenomas (OR, 3.04; 95% CI, 1.04 to 8.88) compared to the patients with excellent bowel preparation.

Conclusions

The risk of missing polyps and adenomas during screening colonoscopy is significantly affected by bowel preparation status. It seems appropriate to shorten the colonoscopy follow-up interval for patients with suboptimal bowel preparation.