Overall Polyp Detection Rate as a Surrogate Measure for Screening Efficacy Independent of Histopathology: Evidence from National Endoscopy Database.

Adenoma detection rate (ADR) is challenging to measure, given its dependency on pathology reporting. Polyp detection rate (PDR) (percentage of screening colonoscopies detecting a polyp) is a proposed alternative to overcome this issue. Overall PDR from all colonoscopies is a relatively novel concept, with no large-scale studies comparing overall PDR with screening-only PDR. The aim of the study was to compare PDR from screening, surveillance, and diagnostic indications with overall PDR and evaluate any correlation between individual endoscopist PDR by indication to determine if overall PDR can be a valuable surrogate for screening PDR. Our study analyzed a prospectively collected national endoscopy database maintained by the National Institute of Health from 2009 to 2014. Out of 354,505 colonoscopies performed between 2009-2014, 298,920 (n = 110,794 average-risk screening, n = 83,556 average-risk surveillance, n = 104,770 diagnostic) met inclusion criteria. The median screening PDR was 25.45 (IQR 13.15-39.60), comparable with the median overall PDR of 24.01 (IQR 11.46-35.86, p = 0.21). Median surveillance PDR was higher at 33.73 (IQR 16.92-47.01), and median diagnostic PDR was lower at 19.35 (IQR 9.66-29.17), compared with median overall PDR 24.01 (IQR 11.46-35.86; p < 0.01). The overall PDR showed excellent concordance with screening, surveillance, and diagnostic PDR (r > 0.85, p < 0.01, 2-tailed). The overall PDR is a reliable and pragmatic surrogate for screening PDR and can be measured in real time, irrespective of colonoscopy indication.

Assessment of the role of false-positive alerts in computer-aided polyp detection for assistance capabilities.

BACKGROUND AND AIM: False positives (FPs) pose a significant challenge in the application of artificial intelligence (AI) for polyp detection during colonoscopy. The study aimed to quantitatively evaluate the impact of computer-aided polyp detection (CADe) systems' FPs on endoscopists. METHODS: The model's FPs were categorized into four gradients: 0-5, 5-10, 10-15, and 15-20 FPs per minute (FPPM). Fifty-six colonoscopy videos were collected for a crossover study involving 10 endoscopists. Polyp missed rate (PMR) was set as primary outcome. Subsequently, to further verify the impact of FPPM on the assistance capability of AI in clinical environments, a secondary analysis was conducted on a prospective randomized controlled trial (RCT) from Renmin Hospital of Wuhan University in China from July 1 to October 15, 2020, with the adenoma detection rate (ADR) as primary outcome. RESULTS: Compared with routine group, CADe reduced PMR when FPPM was less than 5. However, with the continuous increase of FPPM, the beneficial effect of CADe gradually weakens. For secondary analysis of RCT, a total of 956 patients were enrolled. In AI-assisted group, ADR is higher when FPPM ≤ 5 compared with FPPM > 5 (CADe group: 27.78% vs 11.90%; P = 0.014; odds ratio [OR], 0.351; 95% confidence interval [CI], 0.152-0.812; COMBO group: 38.40% vs 23.46%, P = 0.029; OR, 0.427; 95% CI, 0.199-0.916). After AI intervention, ADR increased when FPPM ≤ 5 (27.78% vs 14.76%; P = 0.001; OR, 0.399; 95% CI, 0.231-0.690), but no statistically significant difference was found when FPPM > 5 (11.90% vs 14.76%, P = 0.788; OR, 1.111; 95% CI, 0.514-2.403). CONCLUSION: The level of FPs of CADe does affect its effectiveness as an aid to endoscopists, with its best effect when FPPM is less than 5.

The Yield of Colonoscopy in the Evaluation of Constipation: An Age-Based Analysis of Outcome.

Background: Chronic constipation, a prevalent gastrointestinal complaint, exhibits rising incidence and diverse clinical implications, especially among the aging population. This study aims to assess colonoscopy performance in chronic constipation across age groups, comprehensively evaluating diagnostic yield and comparing results with average-risk controls. Methods: A retrospective analysis was conducted on 50,578 colonoscopy procedures performed over 12 years, including 5478 constipated patients. An average-risk control group (n = 4100) was included. Data extracted from electronic medical records covered demographics, operational aspects, and colonoscopy findings. The primary outcome measures included the diagnosis rate of colorectal cancer (CRC), polyp detection rate (PDR), and inflammatory bowel disease (IBD) diagnoses in constipated patients versus controls, with age-based and multivariate analyses. Results: Constipated patients exhibiting lower rates of adequate bowel preparation (92.7% vs. 85.3%; p < 0.001) and a lower cecal intubation rate. No significant variances between CRC and PDR were observed between constipated and controls, except for a potential of a slightly elevated CRC risk in constipated patients older than 80 (2.50% vs. 0% in controls; p = 0.07). Multivariate analysis demonstrated, across all age groups, that constipation did not confer an increased risk for CRC or polyp detection. Younger constipated patients exhibited a higher rate of IBD diagnoses (1.7% vs. 0.1% in controls; p < 0.001). Conclusions: Constipation did not confer an increased risk for CRC or polyps, among any age groups, except for a potential signal of elevated CRC risk in patients older than 80; additionally, it was associated with higher rates of IBD in younger patients.

TFCNet: A texture-aware and fine-grained feature compensated polyp detection network.

PURPOSE: Abnormal tissue detection is a prerequisite for medical image analysis and computer-aided diagnosis and treatment. The use of neural networks (CNN) to achieve accurate detection of intestinal polyps is beneficial to the early diagnosis and treatment of colorectal cancer. Currently, image detection models using multi-scale feature processing perform well in polyp detection. However, these methods do not fully consider the misalignment of information in the process of feature scale change, resulting in the loss of fine-grained features, and eventually cause the missed and false detection of targets. METHOD: To solve this problem, a texture-aware and fine-grained feature compensated polyp detection network (TFCNet) is proposed in this paper. Firstly, design Texture Awareness Module (TAM) to excavate the rich texture information from the low-level layers and utilize high-level semantic information for background suppression, thereby capturing purer fine-grained features. Secondly, the Texture Feature Enhancement Module (TFEM) is designed to enhance the low-level texture information in TAM, and the enhanced texture features were fused with the high-level features. By making full use of the low-level texture features and multi-scale context information, the semantic consistency and integrity of the features were ensured. Finally, the Residual Pyramid Splittable Attention Module (RPSA) is designed to balance the loss of channel information caused by skip connections, and further improve the detection performance of the network. RESULTS: Experimental results on 4 datasets demonstrate that the TFCNet network outperforms existing methods. Particularly, on the large dataset PolypSets, the mAP@0.5-0.95 has been improved to 88.9%. On the small datasets CVC-ClinicDB and Kvasir, the mAP@0.5-0.95 is increased by 2% and 1.6%, respectively, compared to the baseline, showcasing a significant superiority over competing methods.

Excellent Bowel Preparation Quality Is Not Superior to Good Bowel Preparation Quality for Improving Adenoma/Polyp Detection Rate.

Background: Adequate bowel preparation quality is essential for high-quality colonoscopy according to the current guidelines. However, the excellent effect of bowel preparation on adenoma/polyp detection rate (ADR/PDR) remained controversial. Methods: During the period from December 2020 to August 2022, a total of 1566 consecutive patients underwent colonoscopy by an endoscopist. Their medical records were reviewed. According to the Boston bowel preparation scale, patients were divided into excellent, good, and poor bowel preparation quality groups. ADR/PDR, diminutive ADR/PDR, small ADR/PDR, intermediate ADR/PDR, large ADR/PDR, and number of adenomas/polyps were compared among them. Logistic regression analyses were performed to identify the factors that were significantly associated with ADR/PDR. Results: Overall, 1232 patients were included, of whom 463, 636, and 133 were assigned to the excellent, good, and poor groups, respectively. The good group had a significantly higher ADR/PDR (63% vs 55%, P = .015) and a larger number of adenomas/polyps (2.5 ± 3.2 vs 2.0 ± 2.8, P = .030) than the poor group. Both ADR/PDR (63% vs 55%, P = .097) and number of adenomas/polyps (2.2 ± 2.8 vs 2.0 ± 2.8, P = .219) were not significantly different between excellent and poor groups. The excellent (9% vs 4%, P = .045) and good (9% vs 4%, P = .040) groups had a significantly higher intermediate ADR/PDR than the poor group. Logistic regression analyses showed that either good (odds ratio [OR] = 1.786, 95% CI = 1.046-3.047, P = .034) or excellent (OR = 2.179, 95% CI = 1.241-3.826, P = .007) bowel preparation quality was independently associated with a higher ADR/PDR compared with poor bowel preparation quality. Excellent (OR = 1.202, 95% CI = 0.848-1.704, P = .302) bowel preparation quality was not independently associated with a higher ADR/PDR compared with good bowel preparation quality. Conclusions: The pursuit of excellence in bowel preparation does not show an association with increased ADR/PDR and number of adenomas/polyps compared with a good level. In addition, our study further contributes to the existing evidence that poor bowel preparation compromises ADR/PDR and number of adenomas/polyps.

Spatio-Temporal Feature Transformation Based Polyp Recognition for Automatic Detection: Higher Accuracy than Novice Endoscopists in Colorectal Polyp Detection and Diagnosis.

BACKGROUND: Artificial intelligence represents an emerging area with promising potential for improving colonoscopy quality. AIMS: To develop a colon polyp detection model using STFT and evaluate its performance through a randomized sample experiment. METHODS: Colonoscopy videos from the Digestive Endoscopy Center of the First Affiliated Hospital of Anhui Medical University, recorded between January 2018 and November 2022, were selected and divided into two datasets. To verify the model's practical application in clinical settings, 1500 colonoscopy images and 1200 polyp images of various sizes were randomly selected from the test set and compared with the STFT model's and endoscopists' recognition results with different years of experience. RESULTS: In the randomized sample trial involving 1500 colonoscopy images, the STFT model demonstrated significantly higher accuracy and specificity compared to endoscopists with low years of experience (0.902 vs. 0.809, 0.898 vs. 0.826, respectively). Moreover, the model's sensitivity was 0.904, which was higher than that of endoscopists with low, medium, or high years of experience (0.80, 0.896, 0.895, respectively), with statistical significance (P < 0.05). In the randomized sample experiment of 1200 polyp images of different sizes, the accuracy of the STFT model was significantly higher than that of endoscopists with low years of experience when the polyp size was ≤ 0.5 cm and 0.6-1.0 cm (0.902 vs. 0.70, 0.953 vs. 0.865, respectively). CONCLUSIONS: The STFT-based colon polyp detection model exhibits high accuracy in detecting polyps in colonoscopy videos, with a particular efficiency in detecting small polyps (≤ 0.5 cm)(0.902 vs. 0.70, P < 0.001).

Effect of Real-Time Computer-Aided Polyp Detection System (ENDO-AID) on Adenoma Detection in Endoscopists-in-Training: A Randomized Trial.

BACKGROUND: The effect of computer-aided polyp detection (CADe) on adenoma detection rate (ADR) among endoscopists-in-training remains unknown. METHODS: We performed a single-blind, parallel-group, randomized controlled trial in Hong Kong between April 2021 and July 2022 (NCT04838951). Eligible subjects undergoing screening/surveillance/diagnostic colonoscopies were randomized 1:1 to receive colonoscopies with CADe (ENDO-AID[OIP-1]) or not (control) during withdrawal. Procedures were performed by endoscopists-in-training with <500 procedures and <3 years' experience. Randomization was stratified by patient age, sex, and endoscopist experience (beginner vs intermediate level, <200 vs 200-500 procedures). Image enhancement and distal attachment devices were disallowed. Subjects with incomplete colonoscopies or inadequate bowel preparation were excluded. Treatment allocation was blinded to outcome assessors. The primary outcome was ADR. Secondary outcomes were ADR for different adenoma sizes and locations, mean number of adenomas, and non-neoplastic resection rate. RESULTS: A total of 386 and 380 subjects were randomized to CADe and control groups, respectively. The overall ADR was significantly higher in the CADe group than in the control group (57.5% vs 44.5%; adjusted relative risk, 1.41; 95% CI, 1.17-1.72; P < .001). The ADRs for <5 mm (40.4% vs 25.0%) and 5- to 10-mm adenomas (36.8% vs 29.2%) were higher in the CADe group. The ADRs were higher in the CADe group in both the right colon (42.0% vs 30.8%) and left colon (34.5% vs 27.6%), but there was no significant difference in advanced ADR. The ADRs were higher in the CADe group among beginner (60.0% vs 41.9%) and intermediate-level (56.5% vs 45.5%) endoscopists. Mean number of adenomas (1.48 vs 0.86) and non-neoplastic resection rate (52.1% vs 35.0%) were higher in the CADe group. CONCLUSIONS: Among endoscopists-in-training, the use of CADe during colonoscopies was associated with increased overall ADR. (ClinicalTrials.gov, Number: NCT04838951).

3 L split-dose polyethylene glycol is superior to 2 L polyethylene glycol in colonoscopic bowel preparation in relatively high-BMI (≥ 24 kg/m2) individuals: a multicenter randomized controlled trial.

BACKGROUND: Whether body mass index (BMI) is a risk factor for poor bowel preparation is controversial, and the optimal bowel preparation regimen for people with a high BMI is unclear. METHODS: We prospectively included 710 individuals with high BMIs (≥ 24 kg/m2) who were scheduled to undergo colonoscopy from January to November 2021 at 7 hospitals. Participants were randomly allocated into 3 L split-dose polyethylene glycol (PEG) group (n=353) and 2 L PEG group (n=357). The primary outcome was the rate of adequate bowel preparation, and the secondary outcomes included Boston Bowel Preparation Scale (BBPS) score, polyp detection rate, cecal intubation rate, and adverse reactions during bowel preparation. Furthermore, we did exploratory subgroup analyses for adequate bowel preparation. RESULTS: After enrollment, 15 individuals didn't undergo colonoscopy, finally 345 participants took 3 L split-dose PEG regimen, and 350 participants took 2 L PEG regimen for colonoscopic bowel preparation. 3 L split-dose PEG regimen was superior to 2 L PEG regimen in the rate of adequate bowel preparation (81.2% vs. 74.9%, P = 0.045), BBPS score (6.71±1.15 vs. 6.37±1.31, P < 0.001), and the rate of polyp detection (62.0% vs. 52.9%, P = 0.015). The cecal intubation rate was similar in both groups (99.7%). Regarding adverse reactions, individuals were more likely to feel nausea in the 3 L PEG group (30.9% vs. 19.3%; P = 0.001); however, the degree was mild. In the subgroup analysis for adequate bowel preparation, 3 L split-dose PEG regimen performed better than 2 L PEG regimen in the overweight (BMI 25-29.9 kg/m2 ) (P = 0.006) and individuals with constipation (P = 0.044), while no significant differences were observed in relatively normal (BMI 24-24.9 kg/m2) (P = 0.593) and obese individuals (BMI ≥ 30 kg/m2) (P = 0.715). CONCLUSIONS: 3 L split-dose PEG regimen is superior to 2 L PEG regimen for colonoscopic Bowel Preparation in relatively high-BMI individuals, especially overweight individuals (BMI 25-29.9 kg/m2 ). TRIAL REGISTRATION: This trial was registered in the Chinese Clinical Trials Registry (ChiCTR2000039068). The date of first registration, 15/10/2020, http://www.chictr.org.cn.

Recent advances in devices and technologies that might prove revolutionary for colonoscopy procedures.

INTRODUCTION: Colorectal cancer (CRC) is the third most common malignancy and second leading cause of cancer-related mortality in the world. Adenoma detection rate (ADR), a quality indicator for colonoscopy, has gained prominence as it is inversely related to CRC incidence and mortality. As such, recent efforts have focused on developing novel colonoscopy devices and technologies to improve ADR. AREAS COVERED: The main objective of this paper is to provide an overview of advancements in the fields of colonoscopy mechanical attachments, artificial intelligence-assisted colonoscopy, and colonoscopy optical enhancements with respect to ADR. We accomplished this by performing a comprehensive search of multiple electronic databases from inception to September 2023. This review is intended to be an introduction to colonoscopy devices and technologies. EXPERT OPINION: Numerous mechanical attachments and optical enhancements have been developed that have the potential to improve ADR and AI has gone from being an inaccessible concept to a feasible means for improving ADR. While these advances are exciting and portend a change in what will be considered standard colonoscopy, they continue to require refinement. Future studies should focus on combining modalities to further improve ADR and exploring the use of these technologies in other facets of colonoscopy.

The Impact of Artificial Intelligence in Improving Polyp and Adenoma Detection Rate During Colonoscopy: Systematic-Review and Meta-Analysis.

INTRODUCTION: Colonoscopy may detect colorectal polyp and facilitate its removal in order to prevent colorectal cancer. However, substantial miss rate for colorectal adenomas detection still occurred during screening colonoscopy procedure. Nowadays, artificial intelligence (AI) have been employed in trials to improve polyp detection rate (PDR) and adenoma detection rate (ADR). Therefore, we would like to determine the impact of AI in increasing PDR and ADR. METHODS: The present study adhered to the guidelines outlined in the Preferred Reporting Items for Systematic Reviews and Meta-analyses 2020 (PRISMA 2020) statement. To identify relevant literature, comprehensive searches were conducted on major scientific databases, including Pubmed, EBSCO-host, and Proquest. The search was limited to articles published up to November 30, 2022. Inclusion criteria for the study encompassed full-text accessibility, articles written in the English language, and randomized controlled trials (RCTs) that reported both ADR and PDR values, comparing conventional diagnostic methods with AI-aided approaches. To synthesize the data, we computed the combined pooled odds ratio (OR) using a random-effects model. This model was chosen due to the expectation of considerable heterogeneity among the selected studies. To evaluate potential publication bias, the Begg's funnel diagram was employed. RESULTS: A total of 13 studies were included in this study. Colonoscopy with AI had significantly higher PDR compared to without AI (pooled OR 1.46, 95% CI 1.13-1.89, p = 0.003) and higher ADR (pooled OR 1.58, 95% CI 1.37-1.82, p < 0.00001). PDR analysis showed moderate heterogeneity between included studies (p = 0.004; I2=63%). Furthermore, ADR analysis showed moderate heterogeneity (p < 0.007; I2 = 57%). Additionally, the funnels plot of ADR and PDR analysis showed an asymmetry plot and low publication bias. CONCLUSION: AI may improve colonoscopy result quality through improving PDR and ADR.

Trend of the polyp and adenoma detection rate by sex and age in asymptomatic average-risk and high-risk individuals undergoing screening colonoscopy, 2012-2019.

Adenoma detection rate (ADR) is an imperative quality measure for colorectal cancer (CRC) screening. This retrospective observational study aimed to determine the trend of polyp detection rate (PDR) and ADR in asymptomatic average- and high-risk participants in different age groups who underwent screening colonoscopy over the seven years from April 2012 to March 2019 in a tertiary gastroenterology referral center of Iran. Of 1676 participants, 51.8 % were men (mean age 52.3 years). The overall PDR and ADR were 22.7 %, and 13.5 %, respectively. Both Polyps and adenomas were more common in age groups 51-59 and ≥60 years in high-risk patients than in the corresponding groups of average-risk patients (p < 0.05). Also, both PDR and ADR were more frequent in men than in women among all studied age groups, but it was statistically significant only for the youngest age group (16.8 % versus 10.5 %, p < 0.05) for PDR and the oldest age group (19.7 % versus 13 %, p < 0.05) for ADR, respectively. The trend of total ADR was upward over 7 years in both average-risk (6.7 % to 13.3 %) and high-risk (9.8 % to 27 %) groups and across all age groups in both sexes. Multivariable logistic regression revealed that high-risk individuals had an elevated risk of adenoma compared with average-risk patients (OR: 1.6, p = 0.006). Substantial variation in thresholds of polyp and adenoma detection by age, sex, and risk categories emphasizes the need for a risk-adapted approach to CRC screening and prevention programs.

Effectiveness of switching endoscopists for repeat surveillance colonoscopy: a retrospective study.

BACKGROUND: Surveillance colonoscopy decreases colorectal cancer mortality; however, lesions are occasionally missed. Although an appropriate surveillance interval is indicated, variations may occur in the methods used, such as scope manipulation or observation. Therefore, individual endoscopists may miss certain areas. This study aimed to verify the effectiveness of performing repeat colonoscopies with a different endoscopist from the initial procedure. METHODS: We retrospectively reviewed a database of 8093 consecutive colonoscopies performed in the Omori Red Cross Hospital from January 1st 2018 to June 30th 2021. Data from repeat total colonoscopies performed within three months were collected to assess missed lesions. The patients were divided into two groups according to whether the two examinations were performed by different endoscopists (group D) or the same endoscopist (group S). The primary outcome in both groups was the missed lesion detection rate (MLDR). RESULTS: Overall, 205 eligible patients were analyzed. In total, 102 and 103 patients were enrolled in groups D and S, respectively. The MLDR was significantly higher in group D (61.8% vs. 31.1%, P < 0.0001). Multivariate logistic regression analysis for the detection of missed lesions identified performance by the different endoscopists (odds ratio, 3.38; 95% CI, 1.81-6.30), and sufficient withdrawal time (> 6 min) (odds ratio, 3.10; 95% CI, 1.12-8.61) as significant variables. CONCLUSIONS: Overall, our study showed a significant improvement in the detection of missed lesions when performed by different endoscopists. When performing repeat colonoscopy, it is desirable that a different endoscopist perform the second colonoscopy. TRIAL REGISTRATION: This study was approved by the Institutional Review Board of the Omori Red Cross Hospital on November 28, 2022 (approval number:22-43).

Advancing Colorectal Cancer Screening: A Comprehensive Systematic Review of Artificial Intelligence (AI)-Assisted Versus Routine Colonoscopy.

Colorectal cancer (CRC) is a rapidly escalating public health concern, which underlines the significance of its early detection and the need for the refinement of current screening methods. In this systematic review, we aimed to analyze the potential advantages and limitations of artificial intelligence (AI)-based computer-aided detection (CADe) systems as compared to routine colonoscopy. This review begins by shedding light on the global prevalence and mortality rates of CRC, highlighting the urgent need for effective screening techniques and early detection of this cancer type. It addresses the problems associated with undetected adenomas and polyps and the subsequent risk of interval CRC following colonoscopy. The incorporation of AI into diagnostics has been studied, specifically the use of CADe systems which are powered by deep learning. The review summarizes the findings from 13 randomized controlled trials (RCTs) (2019-2023), evaluating the impact of CADe on polyp and adenoma detection. The findings from the studies consistently show that CADe is superior to conventional colonoscopy procedures in terms of adenoma detection rate (ADR) and polyp detection rate (PDR), particularly with regard to small and flat lesions which are easily overlooked. The review acknowledges certain limitations of the included studies, such as potential performance bias and geographic limitations. The review ultimately concludes that AI-assisted colonoscopy can reduce missed lesion rates and improve CRC diagnosis. Collaboration between experts and clinicians is key for successful implementation. In summary, this review analyzes recent RCTs on AI-assisted colonoscopy for polyp and adenoma detection. It describes the likely benefits, limitations, and future implications of AI in enhancing colonoscopy procedures and lowering the incidence of CRC. More double-blinded trials and studies among diverse populations from different countries must be conducted to substantiate and expand upon the findings of this review.

Knowledge, perceptions and behaviours of endoscopists towards the use of artificial intelligence-aided colonoscopy.

BACKGROUND: Recent developments in artificial intelligence (AI) systems have enabled advancements in endoscopy. Deep learning systems, using convolutional neural networks, have allowed for real-time AI-aided detection of polyps with higher sensitivity than the average endoscopist. However, not all endoscopists welcome the advent of AI systems. METHODS: We conducted a survey on the knowledge of AI, perceptions of AI in medicine, and behaviours regarding use of AI-aided colonoscopy, in a single centre 2 months after the implementation of Medtronic's GI Genius in colonoscopy. We obtained a response rate of 66.7% (16/24) amongst consultant-grade endoscopists. Fisher's exact test was used to calculate the significance of correlations. RESULTS: Knowledge of AI varied widely amongst endoscopists. Most endoscopists were optimistic about AI's capabilities in performing objective administrative and clinical tasks, but reserved about AI providing personalised, empathetic care. 68.8% (n = 11) of endoscopists agreed or strongly agreed that GI Genius should be used as an adjunct in colonoscopy. In analysing the 31.3% (n = 5) of endoscopists who disagreed or were ambivalent about its use, there was no significant correlation with their knowledge or perceptions of AI, but a significant number did not enjoy using the programme (p-value = 0.0128) and did not think it improved the quality of colonoscopy (p-value = 0.033). CONCLUSIONS: Acceptance of AI-aided colonoscopy systems is more related to the endoscopist's experience with using the programme, rather than general knowledge or perceptions towards AI. Uptake of such systems will rely greatly on how the device is delivered to the end user.

Clinical Validation Benchmark Dataset and Expert Performance Baseline for Colorectal Polyp Localization Methods.

Colorectal cancer is one of the leading death causes worldwide, but, fortunately, early detection highly increases survival rates, with the adenoma detection rate being one surrogate marker for colonoscopy quality. Artificial intelligence and deep learning methods have been applied with great success to improve polyp detection and localization and, therefore, the adenoma detection rate. In this regard, a comparison with clinical experts is required to prove the added value of the systems. Nevertheless, there is no standardized comparison in a laboratory setting before their clinical validation. The ClinExpPICCOLO comprises 65 unedited endoscopic images that represent the clinical setting. They include white light imaging and narrow band imaging, with one third of the images containing a lesion but, differently to another public datasets, the lesion does not appear well-centered in the image. Together with the dataset, an expert clinical performance baseline has been established with the performance of 146 gastroenterologists, who were required to locate the lesions in the selected images. Results shows statistically significant differences between experience groups. Expert gastroenterologists' accuracy was 77.74, while sensitivity and specificity were 86.47 and 74.33, respectively. These values can be established as minimum values for a DL method before performing a clinical trial in the hospital setting.

Gastric polyp detection module based on improved attentional feature fusion.

Gastric cancer is a deadly disease and gastric polyps are at high risk of becoming cancerous. Therefore, the timely detection of gastric polyp is of great importance which can reduce the incidence of gastric cancer effectively. At present, the object detection method based on deep learning is widely used in medical images. However, as the contrast between the background and the polyps is not strong in gastroscopic image, it is difficult to distinguish various sizes of polyps from the background. In this paper, to improve the detection performance metrics of endoscopic gastric polyps, we propose an improved attentional feature fusion module. First, in order to enhance the contrast between the background and the polyps, we propose an attention module that enables the network to make full use of the target location information, it can suppress the interference of the background information and highlight the effective features. Therefore, on the basis of accurate positioning, it can focus on detecting whether the current location is the gastric polyp or background. Then, it is combined with our feature fusion module to form a new attentional feature fusion model that can mitigate the effects caused by semantic differences in the processing of feature fusion, using multi-scale fusion information to obtain more accurate attention weights and improve the detection performance of polyps of different sizes. In this work, we conduct experiments on our own dataset of gastric polyps. Experimental results show that the proposed attentional feature fusion module is better than the common feature fusion module and can improve the situation where polyps are missed or misdetected.

Image acquisition as novel colonoscopic quality indicator: a single-center retrospective study.

Purpose: In order to reduce the incidence and mortality of colorectal cancer, improving the quality of colonoscopy is the top priority. At present, the adenoma detection rate is the most used index to evaluate the quality of colonoscopy. So, we further verified the relevant factors influencing the quality of colonoscopy and found out the novel quality indicators by studying the relationship between the influencing factors and the adenoma detection rate. Materials/methods: The study included 3824 cases of colonoscopy from January to December 2020. We retrospectively recorded the age and sex of the subjects; the number, size, and histological features of lesions; withdrawal time and the number of images acquired during colonoscopy. We analyzed the associated factors affecting adenoma and polyp detection, and verified their effectiveness with both univariate and multivariate logistic regression analyses. Results: Logistic regression analyses showed that gender, age, withdrawal time and the number of images acquired during colonoscopy could serve as independent predictors of adenoma/polyp detection rate. In addition, adenoma detection rate (25.36% vs. 14.29%) and polyp detection rate (53.99% vs. 34.42%) showed a marked increase when the number of images taken during colonoscopy was ≥29 (P<0.001). Conclusions: Gender, age, withdrawal time and the number of images acquired during colonoscopy are influencing factors for the detection of colorectal adenomas and polyps. And we can gain higher adenoma/polyp detection rate when endoscopists capture more colonoscopic images.

Is There a Difference in Adenoma Detection Rates According to Indication? An Experience in a Panamanian Colorectal Cancer Screening Program.

Background: The benefit of colorectal cancer screening in reducing cancer risk and related death is unclear. There are quality measure indicators and multiple factors that affect the performance of a successful colonoscopy. The main objective of our study was to identify if there is a difference in polyp detection rate (PDR) and adenoma detection rate (ADR) according to colonoscopy indication and which factors might be associated. Methods: We conducted a retrospective review of all colonoscopies performed between January 2018 and January 2019, in a tertiary endoscopic center. All patients ≥ 50 years old scheduled for a nonurgent colonoscopy and screening colonoscopy were included. We stratified the total number of colonoscopies into two categories according to the indication: screening vs. non-screening, and then calculated PDR, ADR and serrated polyp detection rate (SDR). We also performed logistic regression model to identify factors associated with detecting polyps and adenomatous polyps. Results: A total of 1,129 and 365 colonoscopies were performed in the non-screening and screening group, respectively. In comparison with the screening group, PDR and ADR were lower for the non-screening group (33% vs. 25%; P = 0.005 and 17% vs. 13%; P = 0.005). SDR was non-significantly lower in the non-screening group when compared with the screening group (11% vs. 9%; P = 0.53 and 22% vs. 13%; P = 0.007). Conclusion: In conclusion, this observational study reported differences in PDR and ADR depending on screening and non-screening indication. These differences could be related to factors related to the endoscopist, time slot allotted for colonoscopy, population background, and external factors.

Computer-aided automated diminutive colonic polyp detection in colonoscopy by using deep machine learning system; first indigenous algorithm developed in India.

BACKGROUND: Colonic polyps can be detected and resected during a colonoscopy before cancer development. However, about 1/4th of the polyps could be missed due to their small size, location or human errors. An artificial intelligence (AI) system can improve polyp detection and reduce colorectal cancer incidence. We are developing an indigenous AI system to detect diminutive polyps in real-life scenarios that can be compatible with any high-definition colonoscopy and endoscopic video- capture software. METHODS: We trained a masked region-based convolutional neural network model to detect and localize colonic polyps. Three independent datasets of colonoscopy videos comprising 1,039 image frames were used and divided into a training dataset of 688 frames and a testing dataset of 351 frames. Of 1,039 image frames, 231 were from real-life colonoscopy videos from our centre. The rest were from publicly available image frames already modified to be directly utilizable for developing the AI system. The image frames of the testing dataset were also augmented by rotating and zooming the images to replicate real-life distortions of images seen during colonoscopy. The AI system was trained to localize the polyp by creating a 'bounding box'. It was then applied to the testing dataset to test its accuracy in detecting polyps automatically. RESULTS: The AI system achieved a mean average precision (equivalent to specificity) of 88.63% for automatic polyp detection. All polyps in the testing were identified by AI, i.e., no false-negative result in the testing dataset (sensitivity of 100%). The mean polyp size in the study was 5 (± 4) mm. The mean processing time per image frame was 96.4 minutes. CONCLUSIONS: This AI system, when applied to real-life colonoscopy images, having wide variations in bowel preparation and small polyp size, can detect colonic polyps with a high degree of accuracy.

Polyps are detected more often in early colonoscopies.

OBJECTIVE: To examine the time variation in polyp detection for colonoscopies performed in a tertiary hospital and to explore independent factors that predict polyp detection rate (PDR). METHODS: Data on all patients who underwent colonoscopy for the diagnostic purpose at our endoscopy center in Zhongnan Hospital of Wuhan University from January 2021 to December 2021 were reviewed. The start time of included colonoscopies for eligible patients was recorded. PDR and polyps detected per colonoscopy (PPC) were calculated. The endoscopists' schedules were classified into full-day and half-day shifts according to their participation in the morning and afternoon colonoscopies. RESULTS: Data on a total of 12116 colonoscopies were analyzed, with a PDR of 38.03% for all the patients and 46.38% for patients ≥50 years. PDR and PPC significantly decreased as the day progressed (both p < .001). For patients ≥50 years, PDR declined below 40% at 13:00-13:59 and 16:00-16:59. The PDR in the morning was higher than that in the afternoon for both half-day (p = .019) and full-day procedures (p < .001). In multivariate analysis, start time, patient gender, age, conscious sedation, and bowel preparation quality significantly predicted PDR (p < .001). CONCLUSIONS: The polyp detection declined as the day progressed. A continuous work schedule resulted in a subpar PDR. Colonoscopies performed in the morning had a higher PDR than that in the afternoon. Patient gender, age, conscious sedation, and bowel preparation quality were identified as the independent predictors of PDR.