Introduction

According to the latest statistics, the global estimate for the year 2020 pointed to over 1.9 million new cases of colorectal cancer (CRC) (10.0%), making it the third most incident tumor among all cancers. Among men, there were approximately 1.1 million new cases, with an estimated risk of 23.40 cases per 100,000 men. Among women, there were 865 thousand new cases, making it the second most frequent tumor, with an incidence rate of 16.20 cases per 100,000 women. The highest incidence rates of CRC in men were observed in central, northern, and southern Europe. Among women, the highest incidences were observed in Oceania and northern Europe.
In terms of mortality in Brazil, in 2020, there were 20,245 deaths from CRC (9.56 per 100,000). Among men, there were 9,889 deaths, corresponding to 9.55 deaths per 100,000 men. Among women, there were 10,356 deaths, corresponding to 9.57 deaths per 100,000 women.
Colonoscopy is considered the gold standard for CRC screening as it allows for both diagnosis and resection of lesions. Observational studies have shown a reduction of up to 77% in the incidence and 29 to 37% in mortality from CRC compared to the group of patients not undergoing prior colonoscopy.
Following successful validation by the Food and Drug Administration (FDA) in 2021, the Computer-Aided Detection (CADe) system was introduced into clinical practice. Subsequently, it became possible to assess its potential benefit in increasing the adenoma detection rate (ADR) and the rate of adenomas per colonoscopy (APC). On the other hand, some potential drawbacks were identified, such as the downgrading of the endoscopist's skills or an increase in examination time.

Aims & Methods

The aim was to evaluate the adenoma detection rate (ADR) in a group of patients undergoing colonoscopies with AI assistance and in a group of patients undergoing the procedure without AI assistance at Hospital Sírio-Libanês, Bela Vista/SP unit.
The sample size calculation was based on a comparison test of two proportions. At a 5% significance level, minimum power of 80%, considering the values obtained from a previous similar study, 254 individuals were required in each group, totaling n=508.
In the first group, examinations were performed using the Fujifilm processor model VP-7000, with a colonoscope model EC-760ZP-V/M. In this group, throughout the examination period, the CADe system remained activated and operational, allowing AI assistance in lesion detection. It was at the discretion of the colonoscopist whether to resect the lesion after the alarm emitted by the CADe.
In the second group, the examinations were performed using the Olympus processor model CV-190, with a colonoscope model CF-HQ190L/I. In this group, no AI assistance was utilized.

Results

The ADR in the entire sample was 37.0%, with a rate of 36.6% among patients whose examinations were performed without the use of AI and 37.4% among those in which the examination was conducted with the use of AI, without statistical difference. The ADR was also stratified by age groups within each gender, showing no significant association with the use of AI in the examination. Within the female gender, a higher ADR was observed in women aged 60 or older compared to those under 60, but without statistical significance (p=0.451).

Conclusion

In this study, after a retrospective analysis of colonoscopies conducted at Hospital Sírio-Libanês, Bela Vista/SP unit, the ADR of examinations performed with AI did not show statistically significant superiority compared to examinations performed without AI.

References

Ferlay J, Colombet M, Soerjomataram I, Parkin DM, Piñeros M, Znaor A, et al. Cancer statistics for the year 2020: An overview. Int J Cancer. 2021;149(4):778-89.
Dilmaghani S, Coelho-Prabhu N. Role of artificial intelligence in colonoscopy: a literature review of the past, present, and future directions. Tech Innovations Gastrointest Endoscopy. 2023;25(3):399-412.
Hassan C, Wallace MB, Sharma P, Maselli R, Craviotto V, Spadaccini M, et al. New artificial intelligence system: first validation study versus experienced endoscopists for colorectal polyp detection. Gut. 2020;69(5):799-800.
Mori Y, Kudo SE, Misawa M, Saito Y, Ikematsu H, Hotta K, et al. Real-Time Use of Artificial Intelligence in Identification of Diminutive Polyps During Colonoscopy: A Prospective Study. Ann Intern Med. 2018;169(6):357-66.21.
Wang P, Berzin TM, Glissen Brown JR, Bharadwaj S, Becq A, Xiao X, et al. Real-time automatic detection system increases colonoscopic polyp and adenoma detection rates: a prospective randomised controlled study. Gut. 2019;68(10):1813-9.22.
Repici A, Badalamenti M, Maselli R, Correale L, Radaelli F, Rondonotti E, et al. Efficacy of Real-Time Computer-Aided Detection of Colorectal Neoplasia in a Randomized Trial. Gastroenterology. 2020;159:512-20.e7.23.
Byrne MF, Shahidi N, Rex DK. Will computer-aided detection and diagnosis revolutionize colonoscopy. Gastroenterology. 2017;153:1460-4.e1.

Introduction

One-fourth of colorectal neoplasia is missed at screening colonoscopy, representing the leading cause of interval colorectal cancer (I-CRC). Colonoscopy assisted by artificial intelligence (AI) with computer-aided polyp detection (CADe) showed higher adenoma detection rate (ADR) and higher adenoma per colonoscopy compared to white light (WL) colonoscopy in randomized clinical trials (RCTs). Despite ADR is a reliable quality indicator and has been associated to the risk of I-CRC and death, an adequate ADR does not eliminate the risk of lesion miss rate since the relationship between ADR and adenoma miss rate (AMR) is only indirect.

Aims & Methods

The aim of this meta-analysis is to summarise the performance of CADe systems on miss rate of colonic lesions. PubMed/Medline, Embase and Cochrane Library were systematically searched through October 2022 by two independent reviewers (M.M. and A.F.) for tandem-design RCTs comparing miss rate of colonic lesions in CAD-first colonoscopy followed by WL colonoscopy (experimental arm) vs WL-first colonoscopy followed by CAD colonoscopy (control arm).
The primary outcomes were pooled AMR and polyp miss rate (PMR). Secondary outcomes were advanced adenoma miss rate (aAMR) and sessile serrated lesion miss rate (SMR). Lesion miss rate was defined as the number of lesions detected at second colonoscopy divided by the total number of same lesions detected at first and second colonoscopy. A random-effect model was applied for pooling results. Heterogeneity was expressed as I².

Results

Overall, 4 RCTs (1166 patients), were included in the analysis. Pooled AMR was significantly lower in the CAD-first compared to WL-first arm (131/842, 15.5% vs 300/857, 35.0%; RR, 0.45; 95% CI, 0.35-0.58; P<0.001; I²=49%). PMR was also lower in the CAD-first compared to WL-first arm (192/1198, 16.0% vs 474/1244, 38.1%; RR, 0.43; 95% CI, 0.30-0.61; P<0.001; I²=82%).
No significant difference in aAMR (2/23, 8.7% vs 5/32, 15.6%; RR, 1.28; 95% CI, 0.34-4.83; P=0.71; I²=0%) and in SMR (5/43, 11.6% vs 27/67, 40.3%; RR, 0.44; 95% CI, 0.15-1.28; P=0.13; I²=46%) were observed.

Table 1. Characteristics of the studies included in the meta-analysis.

Conclusion

According to available evidence, the use of colonoscopy assisted by AI with CADe results in a significant lower AMR and PMR, with a potential implication in reducing the incidence of I-CRC. No significant difference in aAMR and SMR has been observed, even if these results should be further confirmed in studies with larger sample sizes.

Introduction

There is a growing interest in the role of artificial intelligence in colonoscopy. In this systematic review and meta-analysis, we aimed to evaluate the efficacy of computer-aided detection (CADe) of colorectal adenomas and polyps.

Aims & Methods

We searched MEDLINE, EMBASE and CENTRAL (from inception to December 2022) for randomized controlled trials (RCTs) comparing colonoscopy with CADe versus standard colonoscopy (SC). We performed a random-effects meta-analysis and reported the results as relative risks (RR) or mean difference (MD) with 95% confidence intervals (CI).

Results

Twelve RCTs comprising 11,340 patients were included in the final analysis. The pooled adenoma detection rate (ADR) was significantly higher in the CADe group compared with the SC group (41.4% vs 33%; RR 1.26, 95% CI 1.18 – 1.35). CADe increased the detection of adenomas regardless of their size (≤5mm [RR 1.56, 95% CI 1.38 – 1.77], 6 – 9mm [RR 1.24 95% CI 1.05 – 1.47] and ≥10mm [RR 1.30, 95% CI 1.11 – 1.53]), location (proximal colon [RR 1.41, 95% CI 1.26 – 1.58] and distal colon [RR 1.44, 95% 1.29 – 1.61]) or morphology (polypoid [RR 1.35, 95% CI 1.17 – 1.56] and non-polypoid [RR 1.55, 95% 1.25 – 1.93]). There was no difference between the CADe and SC groups in detecting advanced adenomas or sessile serrated lesions. Colonoscopy withdrawal time was longer in the CADe group compared with the SC group (MD 0.34 minutes, 95% CI 0.17 – 0.51).

Conclusion

Using CADe during colonoscopy is associated with a significant increase in ADR and adenoma per colonoscopy, which was mainly due to the increased detection of diminutive adenomas. Long-term longitudinal studies are required to evaluate the effect of CADe on the risk of interval colorectal cancer and cancer-related deaths.

Introduction

Recently, artificial intelligence-based computer-aided detection (CADe) for colonoscopy has garnered attention due to its effectiveness in increasing the adenoma detection rate (ADR). However, theoretically, CADe systems cannot reduce exposure errors for polyps obscured by objects. In this regard, computer-aided quality improvement (CAQ) which expects to induce the meticulous mucosal inspection and leading to a reduction of blind spot area. In this study, we developed and evaluated CAQ for colonoscopy.

Aims & Methods

This was a single-center, retrospective study conducted from April 1, 2022 to October 31, 2023. We randomly extracted videos of the entire colonoscopies and created 18,922 training images. Quadrant images were made from one image and three endoscopists annotated the following scores; i) (overall) quality of image(not evaluable 0 or evaluable 1), ii) appropriate distance between the colon mucosa and the endoscope(Scored on a scale of 0 to 8), iii) the amount of residue in the colon,(Scored on a scale of 0 to 8) and ⅳ) presence of blind area in the field of view (e.g., fold or flexure) (Scored on a scale of 0 to 4). We conducted supervised machine learning using annotated images. The CAQ deployed multi-head deep convolutional neural network model which can output multiclass prediction from one image. Consequently, the CAQ system outputs aforementioned score from colonoscopy video. To validate its performance, we randomly selected videos from the cecum to the hepatic flexure performed by 19 endoscopists and calculated CAQ model scores. We compared CAQ scores between endoscopists with low and high ADR (ADR cutoff is 35%) and with and without the history of detecting early depressed-type colorectal cancer by endoscopists.

Results

62 videos were included as validation samples. Among the 19 endoscopists, 11 had a low ADR and 8 had a high ADR. The results showed a significant difference in the rate of evaluable images between endoscopists with low and high ADR (median 70% vs. 77%, p = 0.003, t-test). There was also a significant difference in the evaluation score of the distance (median 3.8 vs. 4.2, p = 0.003, t-test). A significant difference was also obtained in the score for the amount of residue in the colon(median 5.3 vs. 5.7, p = 0.01, t-test). However, there was no significant difference in the scores evaluating the presence of blind areas(median 1.01 vs. 1.10, p = 0.13, t-test)
Among the 19 endoscopists, 5 endoscopists had a history of detecting early depressed-type colorectal cancer, while 14 did not. The results showed a significant difference was observed in the rate of evaluable images between endoscopists with and without a history of detecting early depressed-type colorectal cancer(median 80% vs. 71%, p < 0.001, t-test) . Additionally, there was a difference in the evaluation scores for the distance (median 4.5 vs. 3.8, p < 0.001, t-test). There was also a difference in the scores assessing the presence or absence of blind spots (median 1.2 vs. 1.0, p = 0.006, t-test). Moreover, a significant difference was found in the evaluation scores for the amount of residue in the colon(median 5.9 vs. 5.3, p < 0.001, t-test) .

Conclusion

The CAQ system developed in this study was able to quantify the skill of endoscopists. Such CAQ is expected to standardize the observation technique of colonoscopy and may lead to elucidation of detection techniques for significant lesions such as early depressed-type colorectal cancer.