Comparison of Endoscopic and Artificial Intelligence Diagnoses for Predicting the Histological Healing of Ulcerative Colitis in a Real-World Clinical Setting

Background

Artificial intelligence (AI)-assisted colonoscopy systems with contact microscopy capabilities have been reported previously; however, no studies regarding the clinical use of a commercially available system in patients with ulcerative colitis (UC) have been reported. In this study, the diagnostic performance of an AI-assisted ultra-magnifying colonoscopy system for histological healing was compared with that of conventional light non-magnifying endoscopic evaluation in patients with UC.

Methods

The data of 52 patients with UC were retrospectively analyzed. The Mayo endoscopic score (MES) was determined by 3 endoscopists. Using the AI system, healing of the same spot assessed via MES was defined as a predicted Geboes score (GS) < 3.1. The GS was then determined using pathology specimens from the same site.

Results

A total of 191 sites were evaluated, including 159 with a GS < 3.1. The MES diagnosis identified 130 sites as MES0. A total of 120 sites were determined to have healed based on AI. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of MES0 for the diagnosis of GS < 3.1 were 79.2%, 90.6%, 97.7%, 46.8%, and 81.2%, respectively. The AI system performed similarly to MES for the diagnosis of GS < 3.1: sensitivity, 74.2%; specificity: 93.8%; PPV: 98.3%; NPV: 42.3%; and accuracy: 77.5%. The AI system also significantly identified a GS of < 3.1 in the setting of MES1 (P = .0169).

Conclusions

The histological diagnostic yield the MES- and AI-assisted diagnoses was comparable. Healing decisions using AI may avoid the need for histological examinations.

Impact and assessment of training models in interventional endoscopic ultrasound

Interventional endoscopic ultrasound (IEUS) has gained significant popularity in recent years because of its diagnostic and therapeutic capabilities. The proper training of endoscopists is critical to ensure safe and effective procedures. This review study aims to assess the impact of different training models on the competence of trainees performing IEUS. Eight studies that evaluated simulators for IEUS were identified in the medical literature. Various training models have been used, including the EASIE-R, Mumbai EUS, EUS Magic Box, EndoSim, Thai Association for Gastrointestinal Endoscopy model, and an ex vivo porcine model (HiFi SAM). The trainees underwent traditional didactic lectures, hands-on training using simulators, and direct supervision by experienced endoscopists. The effectiveness of these models has been evaluated based on objective and subjective parameters such as technical proficiency, operative time, diagnostic success, and participant feedback. As expected, the majority of skills were improved after the training sessions concluded, although the risk of bias is high in the absence of external validation. It is difficult to determine the ideal simulator among the existing ones because of the wide variation between them in terms of costs, reusability, design, fidelity of anatomical structures and feedback, and types of procedures performed. There is a need for a standardized approach for the evaluation of IEUS simulators and the ways skills are acquired by trainees, as well as a clearer definition of the key personal attributes necessary for developing a physician into a skilled endoscopist capable of performing basic and advanced therapeutic EUS interventions.

Using a Deep Learning Model to Address Interobserver Variability in the Evaluation of Ulcerative Colitis (UC) Severity

The use of endoscopic images for the accurate assessment of ulcerative colitis (UC) severity is crucial to determining appropriate treatment. However, experts may interpret these images differently, leading to inconsistent diagnoses. This study aims to address the issue by introducing a standardization method based on deep learning. We collected 254 rectal endoscopic images from 115 patients with UC, and five experts in endoscopic image interpretation assigned classification labels based on the Ulcerative Colitis Endoscopic Index of Severity (UCEIS) scoring system. Interobserver variance analysis of the five experts yielded an intraclass correlation coefficient of 0.8431 for UCEIS scores and a kappa coefficient of 0.4916 when the UCEIS scores were transformed into UC severity measures. To establish a consensus, we created a model that considered only the images and labels on which more than half of the experts agreed. This consensus model achieved an accuracy of 0.94 when tested with 50 images. Compared with models trained from individual expert labels, the consensus model demonstrated the most reliable prediction results.

Diagnostic accuracy of convolutional neural network-based machine learning algorithms in endoscopic severity prediction of ulcerative colitis: a systematic review and meta-analysis

BACKGROUND AND AIMS

Endoscopic assessment of ulcerative colitis (UC) can be performed by using the Mayo Endoscopic Score (MES) or the Ulcerative Colitis Endoscopic Index of Severity (UCEIS). In this meta-analysis, we assessed the pooled diagnostic accuracy parameters of deep machine learning by means of convolutional neural network (CNN) algorithms in predicting UC severity on endoscopic images.

METHODS

Databases including MEDLINE, Scopus, and Embase were searched in June 2022. Outcomes of interest were the pooled accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Standard meta-analysis methods used the random-effects model, and heterogeneity was assessed using the I2statistics.

RESULTS

Twelve studies were included in the final analysis. The pooled diagnostic parameters of CNN-based machine learning algorithms in endoscopic severity assessment of UC were as follows: accuracy 91.5% (95% confidence interval [CI], 88.3-93.8; I2 = 84%), sensitivity 82.8% (95% CI, 78.3-86.5; I2 = 89%), specificity 92.4% (95% CI, 89.4-94.6; I2 = 84%), PPV 86.6% (95% CI, 82.3-90; I2 = 89%), and NPV 88.6% (95% CI, 85.7-91; I2 = 78%). Subgroup analysis revealed significantly better sensitivity and PPV with the UCEIS scoring system compared with the MES (93.6% [95% CI, 87.5-96.8; I2 = 77%] vs 82% [95% CI, 75.6-87; I2 = 89%], P = .003, and 93.6% [95% CI, 88.7-96.4; I2 = 68%] vs 83.6% [95% CI, 76.8-88.8; I2 = 77%], P = .007, respectively).

CONCLUSIONS

CNN-based machine learning algorithms demonstrated excellent pooled diagnostic accuracy parameters in the endoscopic severity assessment of UC. Using UCEIS scores in CNN training might offer better results than the MES. Further studies are warranted to establish these findings in real clinical settings.

Machine learning for detection and classification of oral potentially malignant disorders: A conceptual review

Oral potentially malignant disorders represent precursor lesions that may undergo malignant transformation to oral cancer. There are many known risk factors associated with the development of oral potentially malignant disorders, and contribute to the risk of malignant transformation. Although many advances have been reported to understand the biological behavior of oral potentially malignant disorders, their clinical features that indicate the characteristics of malignant transformation are not well established. Early diagnosis of malignancy is the most important factor to improve patients' prognosis. The integration of machine learning into routine diagnosis has recently emerged as an adjunct to aid clinical examination. Increased performances of artificial intelligence AI-assisted medical devices are claimed to exceed the human capability in the clinical detection of early cancer. Therefore, the aim of this narrative review is to introduce artificial intelligence terminology, concepts, and models currently used in oncology to familiarize oral medicine scientists with the language skills, best research practices, and knowledge for developing machine learning models applied to the clinical detection of oral potentially malignant disorders.

Beyond colonoscopy: Physical activity as a viable adjunct to prevent colorectal cancer

Colorectal cancer (CRC) is a common cancer with an increasing incidence worldwide. The implementation of a mass screening program has been proven effective in reducing the global burden of CRC, but its effectiveness is not ideal and some metabolic derangements and lifestyle factors were reported to be attributable for such a deficit. Implementing positive lifestyle intervention as primary prevention therefore becomes critical because colorectal carcinogenesis can be promoted by several lifestyle factors, such as a lack of physical activity. Herein, we review the current evidence on the association and possible mechanisms between physical activity and CRC carcinogenesis. In addition, since CRC prevention heavily relies on resection of precancerous polyps and subsequent surveillance by colonoscopy, this review will also explore the impact of physical activity on populations with different colorectal polyp risks and its potential adjunct role in altering surveillance outcomes.

The Link between Food Environment and Colorectal Cancer: A Systematic Review

Food and diet are critical risk factors for colorectal cancer (CRC). Food environments (FEs) can contribute to disease risk, including CRC. This review investigated the link between FEs and CRC incidence and mortality risk. The systematic search of studies utilised three primary journal databases: PubMed, Scopus, and Web of Science. Retrieved citations were screened and the data were extracted from articles related to the FE-exposed populations who were at risk for CRC and death. We evaluated ecological studies and cohort studies with quality assessment and the Newcastle-Ottawa Quality Assessment Form for Cohort Studies, respectively. A descriptive synthesis of the included studies was performed. Out of 89 articles identified, eight were eligible for the final review. The included studies comprised six ecological studies and two cohort studies published from 2013 to 2021. Six articles were from the US, one was from Africa, and one was from Switzerland. All eight studies were of good quality. The significant finding was that CRC incidence was associated with the availability of specific foods such as red meat, meat, animal fats, energy from animal sources, and an unhealthy FE. Increased CRC mortality was linked with the availability of animal fat, red meat, alcoholic beverages, and calorie food availability, residence in food deserts, and lower FE index. There were a variety of associations between CRC and the FE. The availability of specific foods, unhealthy FE, and food desserts impact CRC incidence and mortality. Creating a healthy FE in the future will require focus and thorough planning.

Associations of red and processed meat intake with screen-detected colorectal lesions

Limited data exist regarding the role of meat consumption in early-stage colorectal carcinogenesis. We examined associations of red and processed meat intake with screen-detected colorectal lesions in immunochemical fecal occult blood test (FIT)-positive participants, enrolled in the Norwegian CRCbiome study during 2017-2021, aged 55-77 years. Absolute and energy-adjusted intakes of red and processed meat (combined and individually) were assessed using a validated, semi-quantitative FFQ. Associations between meat intake and screen-detected colorectal lesions were examined using multinomial logistic regression analyses with adjustment for key covariates. Of 1162 participants, 319 presented with advanced colorectal lesions at colonoscopy. High v. low energy-adjusted intakes of red and processed meat combined, as well as red meat alone, were borderline to significantly positively associated with advanced colorectal lesions (OR of 1·24 (95 % CI 0·98, 1·57) and 1·34 (95 % CI 1·07, 1·69), respectively). A significant dose-response relationship was also observed for absolute intake levels (OR of 1·32 (95 % CI 1·09, 1·60) per 100 g/d increase in red and processed meat). For processed meat, no association was observed between energy-adjusted intakes and advanced colorectal lesions. A significant positive association was, however, observed for participants with absolute intake levels ≥ 100 v. < 50 g/d (OR of 1·19 (95 % CI 1·09, 1·31)). In summary, high intakes of red and processed meat were associated with presence of advanced colorectal lesions at colonoscopy in FIT-positive participants. The study demonstrates a potential role of dietary data to improve the performance of FIT-based screening.

Combined Effect of Healthy Lifestyle Factors and Risks of Colorectal Adenoma, Colorectal Cancer, and Colorectal Cancer Mortality: Systematic Review and Meta-Analysis

Background

In addition to adiposity, lifestyle factors such as poor diet, low physical activity, alcohol intake and smoking are noted to be associated with the development of colorectal cancer (CRC). This study aims to investigate the association and dose-response relationship between adherence to a healthy lifestyle and CRC risk.

Methods

A systematic literature search was conducted in MEDLINE and EMBASE for studies examining multiple lifestyle factors with risk of CRC, incident colorectal adenoma (CRA), and CRC-specific mortality through June 2021 without restrictions on language or study design. Meta-analysis was performed to pool hazard ratios using random-effects model. Subgroup analyses were performed based upon study and sample characteristics. Random-effects dose-response analysis was also conducted for CRC risk to assess the effect of each additional healthy lifestyle factor.

Results

A total of 28 studies (18 cohort studies, eight case-control studies, and two cross-sectional study) were included. When comparing subjects with the healthiest lifestyle to those with the least healthy lifestyle, the pooled HR was statistically significant for CRC (0.52, 95% CI 0.44-0.63), colon cancer (0.54, 95% CI 0.44-0.67), rectal cancer (0.51, 95% CI 0.37-0.70), CRA (0.39, 95% CI 0.29-0.53), and CRC-specific mortality (0.65, 95% CI 0.52-0.81). The pooled HR for CRC was 0.91 (95% CI: 0.88-0.94) for each increase in the number of healthy lifestyles. The inverse association between healthy lifestyle and CRC risk was consistently observed in all subgroups (HR ranging from 0.26 to 0.86).

Conclusions

Adoption of a higher number of healthy lifestyles is associated with lower risk of CRC, CRA, and CRC-specific mortality. Promoting healthy lifestyle could reduce the burden of CRC.

Systematic Review Registration

https://www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=231398, identifier CRD42021231398.

Software Analysis of Colonoscopy Videos Enhances Teaching and Quality Metrics

Purpose

Machine learning algorithms were hypothesized as being able to predict the quality of colonoscopy luminal images. This is to enhance training and quality indicators in endoscopy.

Methods

A separate study involving a randomized controlled trial of capped vs. un-capped colonoscopies provided the colonoscopy videos for this study. Videos were analyzed with an algorithm devised by the Australian Institute for Machine Learning. The image analysis validated focus measure, steerable filters-based metrics (SFIL), was used to assess luminal visualization quality and was compared with two independent clinician assessments (C1 and C2). Goodman and Kruskal's gamma (G) measure was used to assess rank correlation data using IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY).

Results

A total of 500 random colonoscopy video clips were extracted and analyzed, 88 being excluded. SFIL scores matched with C1 in 45% and C2 in 42% of cases, respectively. There was a significant correlation between SFIL and C1 (G = 0.644, p < 0.005) and SFIL and C2 (G = 0.734, p < 0.005).

Conclusion

This study demonstrates that machine learning algorithms can recognize the quality of luminal visualization during colonoscopy. We intend to apply this in the future to enhance colonoscopy training and as a metric for quality assessment.

Healthy lifestyle and the risk of conventional adenomas and serrated polyps: Findings from a large colonoscopy screening population

Evidence on the link between healthy lifestyle and colorectal cancer (CRC) precursors is limited. Our study aimed to examine and compare the associations of healthy lifestyle with CRC precursors in adenoma (AD) -carcinoma and serrated pathways. A total of 24,480 participants including 6,309 ADs, 1,343 serrated polyps (SPs), and 16,828 polyp-free controls were included. A healthy lifestyle score (HLS) was constructed based on five lifestyle factors including cigarette smoking, alcohol drinking, physical activity, diet, and body weight, and categorized into least, slightly, moderately, and most healthy. Multivariable logistic regressions were used to estimate odds ratio (OR) and 95% confidence interval (CI). Inverse dose-response associations between the HLS and risk of ADs were observed (OR per 1 score increment for ADs: 0.82 [95% CI 0.79 - 0.84]; for SPs: 0.73 [95% CI 0.69 - 0.78]), and the association with SPs was more evident than with ADs (OR 0.90, 95% CI 0.85 - 0.96). Compared with participants with the least healthy lifestyle, those with the most healthy lifestyle had 47% lower risk of ADs (OR 0.53, 95% CI 0.47 - 0.59) and 70% lower risk of SPs (OR 0.30, 95% CI 0.23 - 0.39), respectively. These inverse associations were consistent across lesion stage and anatomic subsite and not modified by any stratification factors. The risk advancement periods for the most vs. the least healthy lifestyle were -9.49 years for ADs and -20.69 years for SPs. Our findings help confirm the preventive role of healthy lifestyle in colorectal carcinogenesis.

Lifestyle predictors for inconsistent participation to fecal based colorectal cancer screening

BACKGROUND

Consistent participation in colorectal cancer (CRC) screening with repeated fecal immunochemical test (FIT) is important for the success of the screening program. We investigated whether lifestyle risk factors for CRC were related to inconsistent participation in up to four rounds of FIT-screening.

METHOD

We included data from 3,051 individuals who participated in up to four FIT-screening rounds and returned a lifestyle questionnaire. Using logistic regression analyses, we estimated associations between smoking habits, body mass index (BMI), physical activity, alcohol consumption, diet and a healthy lifestyle score (from least favorable 0 to most favorable 5), and inconsistent participation (i.e. not participating in all rounds of eligible FIT screening invitations).

RESULTS

Altogether 721 (24%) individuals were categorized as inconsistent participants Current smoking and BMI ≥30 kg/m² were associated with inconsistent participation; odds ratios (ORs) and 95% confidence intervals (CIs) were 1.54 (1.21-2.95) and 1.54 (1.20-1.97), respectively. A significant trend towards inconsistent participation by a lower healthy lifestyle score was observed (p < 0.05).

CONCLUSIONS

Lifestyle behaviors were associated with inconsistent participation in FIT-screening. Initiatives aimed at increasing participation rates among those with the unhealthiest lifestyle have a potential to improve the efficiency of screening.

Demographic, endoscopic and histological profile of esophageal cancer at the Gastroenterology Department of Maputo Central Hospital from January 2016 to December 2018

INTRODUCTION

esophageal cancer is a major public health problem in Mozambique. It is the nineth most common cancer worldwide in terms of incidence (604.000 new cases/year), and sixth in overall mortality (544.076 deaths/year). In Mozambique esophageal cancer was the seventh most common cancer in males and the fifth in females between 1991 and 2008.

METHODS

it was done a cross-sectional hospital-based epidemiological study, using secondary demographics endoscopic and pathologic features data. A retrospective analysis of the existing information of patients classified as esophageal cancer diagnosed with upper gastrointestinal endoscopy observed from January 1^st, 2016 to December 31^st, 2018 at the Gastroenterology Service of Maputo Central Hospital. A coding sheet was created a priori, and data analysed in SPSS version 20.

RESULTS

of the 205 cases with complete records where included in the analysis, there was a higher frequency of females with 56.6% (116/205). The average age was 59.5 years with standard deviation of ± 12.9 years. Most of the patients were native of southern Mozambique, with 92.7% (190/205), of which Maputo made up 53.2% (109/205). Regarding race, 99.5% (204/205) were black. The most affected endoscopic location was the middle third with 48.8% (100/205), followed by the lower third with 29.8% (61/205) and the upper third with 21.5% (44/205). Squamous cell carcinoma was the most frequent, with 92.7% (190/205), followed by adenocarcinoma with 4.9% (10/205).

CONCLUSION

due to the high number of observed cases of esophageal cancer, a high degree of clinical suspicion is needed for timely diagnosis and more effective treatment. Updated prevalent studies are needed throughout the country to understand the true impact of esophageal cancer on the Mozambican population.

Associations of Evolutionary-Concordance Diet and Lifestyle Pattern Scores with Incident, Sporadic Colorectal Adenoma in a Pooled Case-Control Study

Differences in diet and lifestyle relative to those of our Paleolithic-era ancestors may explain current high incidences of chronic diseases, including colorectal cancer (CRC), in Westernized countries. Previously reported evolutionary-concordance diet and lifestyle pattern scores, reflecting closeness of diet and lifestyle patterns to those of Paleolithic-era humans, were associated with lower CRC incidence. Separate and joint associations of the scores with colorectal adenoma among men and women are unknown. To address this, we pooled data from three case-control studies of incident, sporadic colorectal adenomas (n = 771 cases, 1,990 controls), used participants' responses to food frequency and lifestyle questionnaires to calculate evolutionary-concordance diet and lifestyle pattern scores, and estimated the scores' associations with adenomas using multivariable unconditional logistic regression. The multivariable-adjusted odds ratios comparing those in the highest relative to the lowest diet and lifestyle score quintiles were 0.84 (95% confidence interval [CI] 0.62, 1.12; P_trend:0.03) and 0.41 (95% CI 0.29, 0.59; P_trend:<0.0001), respectively. The inverse associations were stronger for high-risk adenomas, and among those with both high relative to those with both low diet and lifestyle scores. These results suggest that more evolutionary-concordant diet and lifestyle patterns, separately and jointly, may be associated with lower risk for incident, sporadic colorectal adenoma.Supplemental data for this article is available online at https://doi.org/10.1080/01635581.2021.2002919 .

White-Light Endoscopic Colorectal Lesion Detection Based on Improved YOLOv5

As an effective tool for colorectal lesion detection, it is still difficult to avoid the phenomenon of missed and false detection when using white-light endoscopy. In order to improve the lesion detection rate of colorectal cancer patients, this paper proposes a real-time lesion diagnosis model (YOLOv5x-CG) based on YOLOv5 improvement. In this diagnostic model, colorectal lesions were subdivided into three categories: micropolyps, adenomas, and cancer. In the course of convolutional network training, Mosaic data enhancement strategy was used to improve the detection rate of small target polyps. At the same time, coordinate attention (CA) mechanism was introduced to take into account channel and location information in the network, so as to realize the effective extraction of three kinds of pathological features. The Ghost module was also used to generate more feature maps through linear processing, which reduces the stress of learning model parameters and speeds up detection. The experimental results show that the lesion diagnosis model proposed in this paper has a more rapid and accurate lesion detection ability, and the AP value of polyps, adenomas, and cancer is 0.923, 0.955, and 0.87, and mAP@50 is 0.916.

Development of artificial intelligence technology in diagnosis, treatment, and prognosis of colorectal cancer

Artificial intelligence (AI) technology has made leaps and bounds since its invention. AI technology can be subdivided into many technologies such as machine learning and deep learning. The application scope and prospect of different technologies are also totally different. Currently, AI technologies play a pivotal role in the highly complex and wide-ranging medical field, such as medical image recognition, biotechnology, auxiliary diagnosis, drug research and development, and nutrition. Colorectal cancer (CRC) is a common gastrointestinal cancer that has a high mortality, posing a serious threat to human health. Many CRCs are caused by the malignant transformation of colorectal polyps. Therefore, early diagnosis and treatment are crucial to CRC prognosis. The methods of diagnosing CRC are divided into imaging diagnosis, endoscopy, and pathology diagnosis. Treatment methods are divided into endoscopic treatment, surgical treatment, and drug treatment. AI technology is in the weak era and does not have communication capabilities. Therefore, the current AI technology is mainly used for image recognition and auxiliary analysis without in-depth communication with patients. This article reviews the application of AI in the diagnosis, treatment, and prognosis of CRC and provides the prospects for the broader application of AI in CRC.

Application of Artificial Intelligence to Clinical Practice in Inflammatory Bowel Disease - What the Clinician Needs to Know

Artificial intelligence [AI] techniques are quickly spreading across medicine as an analytical method to tackle challenging clinical questions. What were previously thought of as highly complex data sources, such as images or free text, are now becoming manageable. Novel analytical methods merge the latest developments in information technology infrastructure with advances in computer science. Once primarily associated with Silicon Valley, AI techniques are now making their way into medicine, including in the field of inflammatory bowel diseases [IBD]. Understanding potential applications and limitations of these techniques can be difficult, in particular for busy clinicians. In this article, we explain the basic terminologies and provide a particular focus on the foundations behind state-of-the-art AI methodologies in both imaging and text. We explore the growing applications of AI in medicine, with a specific focus on IBD to inform the practising gastroenterologist and IBD specialist. Finally, we outline possible future uses of these technologies in daily clinical practice.

Evaluation of a risk score based on dietary and lifestyle factors to target a population at risk in colorectal cancer screening

BACKGROUND

The aim of our study was to assess three risk scores to predict lesions, advanced neoplasia (high-risk adenomas and colorectal cancer (CRC)) and CRC in individuals who participate to colorectal cancer screening.

METHODS

The data of dietary and lifestyle risk factors were carried out during 2 mass screening campaigns in France (2013-2016) and the FOBT result was collected until December 2018. The colonoscopy result in positive FOBT was recovered. Three risk scores (Betés score, Kaminski score and adapted-HLI) were calculated to detect individuals at risk of lesions.

RESULTS

The Betés score had an AUROC of 0.63 (95% CI, [0.61-0.66]) for lesions, 0.65 (95% CI, [0.61-0.68]) for advanced neoplasia and 0.65 (95% CI, [0.58-0.72]) for predicting screen-detected CRC. The adapted HLI score had an AUROC of 0.61 (95% CI, [0.58-0.65]) for lesions, 0.61 (95% CI, [0.56-0.65]) for advanced neoplasia and 0.55 (95% CI, [0.45-0.65]) for predicting screen-detected CRC. The Kaminski score had an AUROC of 0.65 (95% CI, [0.63-0.68]) for lesions, 0.65 (95% CI, [0.61-0.68]) for advanced neoplasia and 0.69 (95% CI, [0.62-0.76]) for predicting screen-detected CRC.

CONCLUSION

A simple questionnaire based on CRC risk factors could help general practitioners to identify participants with higher risk of significant colorectal lesions and incite them to perform the fecal occult blood test.

Artificial intelligence in gastrointestinal diseases

Artificial intelligence (AI) applications are growing in medicine. It is important to understand the current state of the AI applications prior to utilizing in disease research and treatment. In this review, AI application in the diagnosis and treatment of gastrointestinal diseases are studied and summarized. In most cases, AI studies had large amounts of data, including images, to learn to distinguish disease characteristics according to a human’s perspectives. The detailed pros and cons of utilizing AI approaches should be investigated in advance to ensure the safe application of AI in medicine. Evidence suggests that the collaborative usage of AI in both diagnosis and treatment of diseases will increase the precision and effectiveness of medicine. Recent progress in genome technology such as genome editing provides a specific example where AI has revealed the diagnostic and therapeutic possibilities of RNA detection and targeting.

Association between colorectal cancer and Fusobacterium nucleatum and Bacteroides fragilis bacteria in Iranian patients: a preliminary study

Background and aim

Recent studies have proposed that commensal bacteria might be involved in the development and progression of gastrointestinal disorders such as colorectal cancer (CRC). Therefore, in this study, the relative abundance of Fusobacterium nucleatum, Bacteroides fragilis, Streptococcus bovis/gallolyticus, and Enteropathogenic Escherichia coli (EPEC) in CRC tissues, and their association with clinicopathologic characteristics of CRC was investigated in Iranian patients. Moreover, the role of these bacteria in the CRC-associated mutations including PIK3CA, KRAS, and BRAF was studied.

Method

To these ends, the noted bacteria were quantified in paired tumors and normal tissue specimens of 30 CRC patients, by TaqMan quantitative Real-Time Polymerase Chain Reaction (qPCR). Next, possible correlations between clinicopathologic factors and mutations in PIK3CA, KRAS, and BRAF genes were analyzed.

Results

In studied samples, B. fragilis was the most abundant bacteria that was detected in 66 and 60% of paired tumor and normal samples, respectively. Furthermore, 15% of the B. fragilis-positive patients were infected with Enterotoxigenic B. fragilis (ETBF) in both adenocarcinoma and matched adjacent normal samples. F. nucleatum was also identified in 23% of tumors and 13% of adjacent normal tissue samples. Moreover, the relative abundance of these bacteria determined by 2^-ΔCT was significantly higher in CRC samples than in adjacent normal mucosa (p < 0.05). On the other hand, our findings indicated that S. gallolyticus and EPEC, compared to adjacent normal mucosa, were not prevalent in CRC tissues. Finally, our results revealed a correlation between F. nucleatum-positive patients and the KRAS mutation (p = 0.02), while analyses did not show any association between bacteria and mutation in PIK3CA and BRAF genes.

Conclusion

The present study is the first report on the analysis of different bacteria in CRC tissue samples of Iranian patients. Our findings revealed that F. nucleatum and B. fragilis might be linked to CRC. However, any link between gut microbiome dysbiosis and CRC remains unknown.

Investigation of the use of microwave ablation with and without cooling urethral perfusion for thermal ablation of the prostate gland in canine cadavers

OBJECTIVE

To investigate the use of microwave ablation (MWA) with cooling urethral perfusion and with no perfusion (MWA-UP and MWA-NP, respectively) for prostate gland ablation in canine cadavers.

ANIMALS

Cadavers of 18 sexually intact male dogs.

PROCEDURES

After technique refinement in 2 cadavers, laparotomy with ultrasound-guided MWA-UP (n = 8) or MWA-NP (8) of the prostate gland was performed in 16 cadavers. Normograde cystourethroscopy was performed before and after treatment; recorded images were reviewed in a blinded manner for scoring of urethral mucosal discoloration and loss of integrity. Difficulty with cystoscope insertion was recorded if present. Excised prostate glands were fixed for serial sectioning, gross measurements, and calculation of percentage ablation. Percentages of prostate tissue necrosis from MWA, denuded urethral mucosa, and depth of epithelial surface loss in an adjacent section of the colon were estimated histologically. Variables of interest were statistically analyzed.

RESULTS

Difficulty with cystoscope insertion after treatment was significantly more common and scores for urethral mucosal discoloration and loss of integrity were significantly higher (indicating more severe lesions) for the MWA-NP group than for the MWA-UP group. The histologically assessed percentage of denuded urethral mucosa was also greater for the MWA-NP group. Overall median percentage prostate gland ablation was 73%; this result was not associated with prostate gland volume or chronological order of treatment.

CONCLUSIONS AND CLINICAL RELEVANCE

MWA-UP induced subtotal thermal necrosis of prostate glands in canine cadavers while limiting urethral mucosal injury. Further study is required to optimize the technique and evaluate its safety and efficacy in vivo as a future curative-intent treatment for prostatic tumors in dogs.

Deep Analysis of the Peripheral Immune System in IBD Reveals New Insight in Disease Subtyping and Response to Monotherapy or Combination Therapy

BACKGROUND

Inflammatory bowel disease (IBD) is a complex disease with variable presentation, progression, and response to therapies. Current disease classification is based on subjective clinical phenotypes. The peripheral blood immunophenome can reflect local inflammation, and thus we measured 39 circulating immune cell types in a large cohort of IBD and control subjects and performed immunotype:phenotype associations.

METHODS

We performed fluorescence-activated cell sorting or CyTOF analysis on blood from 728 Crohn's disease, 464 ulcerative colitis, and 334 non-IBD patients, with available demographics, endoscopic and clinical examinations and medication use.

RESULTS

We observed few immune cell types commonly affected in IBD (lowered natural killer cells, B cells, and CD45RA^- CD8 T cells). Generally, the immunophenome was distinct between ulcerative colitis and Crohn's disease. Within disease subtype, there were further distinctions, with specific immune cell types associating with disease duration, behavior, and location. Thiopurine monotherapy altered abundance of many cell types, often in the same direction as disease association, while anti-tumor necrosis factor (anti-TNF) monotherapy demonstrated an opposing pattern. Concomitant use of an anti-TNF and thiopurine was not synergistic, but rather was additive. For example, thiopurine monotherapy use alone or in combination with anti-TNF was associated with a dramatic reduction in major subclasses of B cells.

CONCLUSIONS

We present a peripheral map of immune cell changes in IBD related to disease entity and therapies as a resource for hypothesis generation. We propose the changes in B cell subsets could affect antibody formation and potentially explain the mechanism behind the superiority of combination therapy through the impact of thiopurines on pharmacokinetics of anti-TNFs.

Selection of patients for large mailed fecal immunochemical test colorectal cancer screening outreach programs: A systematic review

OBJECTIVE

Digital health care offers an opportunity to scale and personalize cancer screening programs, such as mailed outreach for colorectal cancer (CRC) screening. However, studies that describe the patient selection strategy and process for CRC screening are limited. Our objective was to evaluate implementation strategies for selecting patients for CRC screening programs in large health care systems.

METHODS

We conducted a systematic review of 30 studies along with key informant surveys and interviews to describe programmatic implementation strategies for selecting patients for CRC screening. PubMed and Embase were searched since inception through December 2018, and hand searches were performed of the retrieved reference lists but none were incorporated (n = 0). No language exclusions were applied.

RESULTS

Common criteria for outreach exclusion included: being up-to-date with routine CRC screening (n = 22), comorbidities (n = 20), and personal history (n = 22) or family history of cancer (n = 9). Key informant surveys and interviews were performed (n = 28) to understand data sources and practices for patient outreach selection, and found that 13 studies leveraged electronic medical care records, 10 studies leveraged a population registry (national, municipal, community, health), 4 studies required patient opt-in, and 1 study required primary care provider referral. Broad ranges in fecal immunochemical test completion were observed in community clinic (n = 8, 31.0-59.6%), integrated health system (n = 5, 21.2-82.7%), and national regional CRC screening programs (n = 17, 23.0-64.7%). Six studies used technical codes, and four studies required patient self-reporting from a questionnaire to participate.

CONCLUSION

This systematic review provides health systems with the diverse outreach practices and technical tools to support efforts to automate patient selection for CRC screening outreach.

Individual and Joint Associations of Genetic Risk and Healthy Lifestyle Score with Colorectal Neoplasms Among Participants of Screening Colonoscopy

Genetic and lifestyle factors contribute to colorectal cancer risk. We investigated their individual and joint associations with various stages of colorectal carcinogenesis. We assessed associations of a polygenic risk score (PRS) and a healthy lifestyle score (HLS) with presence of nonadvanced adenomas and advanced neoplasms among 2,585 participants of screening colonoscopy from Germany. The PRS and HLS individually showed only weak associations with presence of nonadvanced adenomas; stronger associations were observed with advanced neoplasms (ORs, 95% CI, for highest vs. lowest risk tertile: PRS 2.27, 1.78-2.88; HLS 1.96, 1.53-2.51). The PRS was associated with higher odds of advanced neoplasms among carriers of any neoplasms (1.65, 1.23-2.22). Subjects in the highest risk tertile (vs. lowest tertile) of both scores had higher risks for nonadvanced adenomas (1.77, 1.09-2.86), for advanced neoplasms (3.95, 2.53-6.16) and, among carriers of any neoplasms, for advanced versus nonadvanced neoplasms (2.26, 1.31-3.92). Both scores were individually associated with increased risk of nonadvanced adenomas and, much more pronounced, advanced neoplasms. The similarly strong association in relative terms across all levels of genetic risk implies that a healthy lifestyle may be particularly beneficial in those at highest genetic risk, given that the same relative risk reduction in this group would imply a stronger absolute risk reduction. Genetic factors may be of particular relevance for the transition of nonadvanced to advanced adenomas. PREVENTION RELEVANCE: Genetic factors have strong impact on the risk of colorectal neoplasms, which may be reduced by healthy lifestyle. Similarly strong associations in relative terms across all levels of genetic risk imply that a healthy lifestyle may be beneficial due to higher absolute risk reduction in those at highest genetic risk.

Real-Time Polyp Detection, Localization and Segmentation in Colonoscopy Using Deep Learning

Computer-aided detection, localisation, and segmentation methods can help improve colonoscopy procedures. Even though many methods have been built to tackle automatic detection and segmentation of polyps, benchmarking of state-of-the-art methods still remains an open problem. This is due to the increasing number of researched computer vision methods that can be applied to polyp datasets. Benchmarking of novel methods can provide a direction to the development of automated polyp detection and segmentation tasks. Furthermore, it ensures that the produced results in the community are reproducible and provide a fair comparison of developed methods. In this paper, we benchmark several recent state-of-the-art methods using Kvasir-SEG, an open-access dataset of colonoscopy images for polyp detection, localisation, and segmentation evaluating both method accuracy and speed. Whilst, most methods in literature have competitive performance over accuracy, we show that the proposed ColonSegNet achieved a better trade-off between an average precision of 0.8000 and mean IoU of 0.8100, and the fastest speed of 180 frames per second for the detection and localisation task. Likewise, the proposed ColonSegNet achieved a competitive dice coefficient of 0.8206 and the best average speed of 182.38 frames per second for the segmentation task. Our comprehensive comparison with various state-of-the-art methods reveals the importance of benchmarking the deep learning methods for automated real-time polyp identification and delineations that can potentially transform current clinical practices and minimise miss-detection rates.

Artificial Intelligence in Colorectal Polyp Detection and Characterization

BACKGROUND

Over the past 20 years, the advancement of artificial intelligence (AI) and deep learning (DL) has allowed for fast sorting and analysis of large sets of data. In the field of gastroenterology, colorectal screening procedures produces an abundance of data through video and imaging. With AI and DL, this information can be used to create systems where automatic polyp detection and characterization is possible. Convoluted Neural Networks (CNNs) have proven to be an effective way to increase polyp detection and ultimately adenoma detection rates. Different methods of polyp characterization of being hyperplastic vs. adenomatous or non-neoplastic vs. neoplastic has also been investigated showing promising results.

FINDINGS

The rate of missed polyps on colonoscopy can be as high as 25%. At the beginning of the 2000s, hand-crafted machine learning (ML) algorithms were created and trained retrospectively on colonoscopy images and videos, achieving high sensitivity, specificity, and accuracy of over 90% in many of the studies. Over time, the advancement of DL and CNNs has allowed algorithms to be trained on non-medical images and applied retrospectively to colonoscopy videos and images with similar results. Within the past few years, these algorithms have been applied in real-time colonoscopies and has shown mixed results, one showing no difference while others showing increased polyp detection.Various methods of polyp characterization have also been investigated. Through AI, DL, and CNNs polyps can be identified has hyperplastic/adenomatous or non-neoplastic/neoplastic with high sensitivity, specificity, and accuracy. One of the research areas in polyp characterization is how to capture the polyp image. This paper looks at different modalities of characterizing polyps such as magnifying narrow band imaging (NBI), endocytoscopy, laser-induced florescent spectroscopy, auto-florescent endoscopy, and white-light endoscopy.

CONCLUSIONS

Overall, much progress has been made in automatic detection and characterization of polyps in real time. Barring ethical or mass adoption setbacks, it is inevitable that AI will be involved in the field of GI, especially in colorectal polyp detection and identification.

Interobserver and intraobserver reliability for 2 grading systems for gastric ulcer syndrome in horses

Background

Grading of equine gastric ulcer syndrome (EGUS) is undertaken in clinical and research settings, but the reliability of EGUS grading systems is poorly understood.

Hypothesis/Objectives

Investigate interobserver and intraobserver reliability of an established ordinal grading system and a novel visual analog scale (VAS), and assess the influence of observer experience.

Animals

Sixty deidentified gastroscopy videos.

Methods

Six observers (3 specialists and 3 residents) graded videos using the EGUS Council (EGUC) system and VAS. Observers graded the videos three 3 for each system, using a cross‐over design with at least 1 week between each phase. The order of videos was randomized for each phase.

Methods

Interobserver and intraobserver reliability were estimated using Gwet's agreement coefficient with ordinal weights applied (AC2) for the EGUC system and the intraclass correlation coefficient (ICC) for the VAS.

Results

Using the EGUC system, interobserver reliability was substantial for squamous (AC2 = 0.69; 95% confidence interval [CI], 0.57‐0.80) and glandular mucosa (AC2 = 0.72; 95% CI, 0.70‐0.75), and intraobserver reliability was substantial for squamous (AC2 = 0.80; 95% CI, 0.71‐0.90) and glandular mucosa (AC2 = 0.80; 95% CI, 0.74‐0.86). Interobserver reliability using the VAS was moderate for squamous (ICC = 0.64; 95% CI, 0.31‐0.96) and poor for glandular mucosa (ICC = 0.35; 95% CI, 0.06‐0.64), and intraobserver reliability was moderate for squamous (ICC = 0.74; 95% CI, 0.62‐0.86) and glandular mucosa (ICC = 0.56; 95% CI, 0.39‐0.72).

Conclusions and Clinical Importance

The EGUC system had acceptable intraobserver and interobserver reliability and performed well regardless of observer experience. Familiarity and observer experience improved reliability of the VAS.

Artificial intelligence-based diagnostic system classifying gastric cancers and ulcers: comparison between the original and newly developed systems

BACKGROUND

We previously reported for the first time the usefulness of artificial intelligence (AI) systems in detecting gastric cancers. However, the "original convolutional neural network (O-CNN)" employed in the previous study had a relatively low positive predictive value (PPV). Therefore, we aimed to develop an advanced AI-based diagnostic system and evaluate its applicability for the classification of gastric cancers and gastric ulcers.

METHODS

We constructed an "advanced CNN" (A-CNN) by adding a new training dataset (4453 gastric ulcer images from 1172 lesions) to the O-CNN, which had been trained using 13 584 gastric cancer and 373 gastric ulcer images. The diagnostic performance of the A-CNN in terms of classifying gastric cancers and ulcers was retrospectively evaluated using an independent validation dataset (739 images from 100 early gastric cancers and 720 images from 120 gastric ulcers) and compared with that of the O-CNN by estimating the overall classification accuracy.

RESULTS

The sensitivity, specificity, and PPV of the A-CNN in classifying gastric cancer at the lesion level were 99.0 % (95 % confidence interval [CI] 94.6 %-100 %), 93.3 % (95 %CI 87.3 %-97.1 %), and 92.5 % (95 %CI 85.8 %-96.7 %), respectively, and for classifying gastric ulcers were 93.3 % (95 %CI 87.3 %-97.1 %), 99.0 % (95 %CI 94.6 %-100 %), and 99.1 % (95 %CI 95.2 %-100 %), respectively. At the lesion level, the overall accuracies of the O- and A-CNN for classifying gastric cancers and gastric ulcers were 45.9 % (gastric cancers 100 %, gastric ulcers 0.8 %) and 95.9 % (gastric cancers 99.0 %, gastric ulcers 93.3 %), respectively.

CONCLUSION

The newly developed AI-based diagnostic system can effectively classify gastric cancers and gastric ulcers.

Clinical adoption of robotics in endoscopy: Challenges and solutions

The endoscope was traditionally used as a diagnostic instrument. In past decades, it has increasingly been adapted for therapeutic intents. Subsequently, the master–slave robotic concept was introduced into the field of endoscopy to potentially reduce the difficulty and complication rates of endoscopic therapeutic procedures. As interest in robotic endoscopy intensified, progressively more robotic endoscopic platforms were developed, tested, and introduced. Nevertheless, the future of robotic endoscopy hinges on the ability to meet specific clinical needs of procedurists. Three aspects are vital in ensuring continued success and clinical adoption of the robotic endoscope—demonstration of clinical safety and cost‐efficacy of the device, widespread availability of directed training opportunities to enhance technical skills and clinical decision‐making capabilities of the procedurist, and continued identification of new clinical applications beyond the current uses of the device. This review provides a brief discussion of the historical development of robotic endoscopy, current robotic endoscopic platforms, use of robotic endoscopy in conventional therapeutic endoscopic procedures, and the future of robotic endoscopy.

A primer on artificial intelligence and its application to endoscopy

Artificial intelligence (AI) has emerged as a powerful and exciting new technology poised to impact many aspects of health care. In endoscopy, AI is now being used to detect and characterize benign and malignant GI lesions and assess malignant lesion depth of invasion. It will undoubtedly also find use in capsule endoscopy and inflammatory bowel disease. Herein, we provide the general endoscopist with a brief overview of AI and its emerging uses in our field. We also touch on the challenges of incorporating AI into clinical practice, such as workflow integration, data storage, and data privacy.

Artificial intelligence-assisted esophageal cancer management: Now and future

Esophageal cancer poses diagnostic, therapeutic and economic burdens in high-risk regions. Artificial intelligence (AI) has been developed for diagnosis and outcome prediction using various features, including clinicopathologic, radiologic, and genetic variables, which can achieve inspiring results. One of the most recent tasks of AI is to use state-of-the-art deep learning technique to detect both early esophageal squamous cell carcinoma and esophageal adenocarcinoma in Barrett’s esophagus. In this review, we aim to provide a comprehensive overview of the ways in which AI may help physicians diagnose advanced cancer and make clinical decisions based on predicted outcomes, and combine the endoscopic images to detect precancerous lesions or early cancer. Pertinent studies conducted in recent two years have surged in numbers, with large datasets and external validation from multi-centers, and have partly achieved intriguing results of expert’s performance of AI in real time. Improved pre-trained computer-aided diagnosis algorithms in the future studies with larger training and external validation datasets, aiming at real-time video processing, are imperative to produce a diagnostic efficacy similar to or even superior to experienced endoscopists. Meanwhile, supervised randomized controlled trials in real clinical practice are highly essential for a solid conclusion, which meets patient-centered satisfaction. Notably, ethical and legal issues regarding the black-box nature of computer algorithms should be addressed, for both clinicians and regulators.

Do You See What I See? An Assessment of Endoscopic Lesions Recognition and Description by Gastroenterology Trainees and Staff Physicians

Background

Gastroenterologists should accurately describe endoscopic findings and integrate them into management plans. We aimed to determine if trainees and staff are describing inflammatory bowel disease (IBD) lesions in a similar manner.

Methods

Using 20 ileocolonoscopy images, participants described IBD inflammatory burden based on physician severity rating, and Mayo endoscopic score (MES) (ulcerative colitis [UC]) or simple endoscopic score (SES-CD) (Crohn’s disease [CD]). Images were selected based on agreement by three IBD experts. Findings of varying severity were presented; 10 images included a question about management. We examined inter-observer agreement among trainees and staff, compared trainees to staff, and determined accuracy of response comparing both groups to IBD experts.

Results

One hundred and twenty-nine staff and 47 trainees participated from across Canada. There was moderate inter-rater agreement using physician severity rating (κ = 0.53 UC and 0.52 CD for staff, κ = 0.51 UC and 0.43 CD for trainees). There was moderate inter-rater agreement for MES for staff and trainees (κ = 0.49 and 0.48, respectively), but fair agreement for SES-CD (κ = 0.37 and 0.32, respectively). For accuracy of response, the mean score was 68.7% for staff and 63.7% for trainees (P = 0.028). Both groups identified healed bowel or severe disease better than mild/moderate (P < 0.05). There was high accuracy for management, but staff scored higher than trainees for UC (P < 0.01).

Conclusion

Inter-rater agreement on description of IBD lesions was moderate at best. Staff and trainees more accurately describe healed and severe disease, and better describe lesions in UC than CD.

Do Histologic Features Help Predict Colectomy in Pediatric Patients Presenting With Acute Severe Colitis?

BACKGROUND

Multiple prior studies have looked at clinical and laboratory parameters in ulcerative colitis to predict prognosis, but individual histologic features of inflammation and their prognostic significance have not been well studied. The purpose of our study was to determine whether histologic features at presentation with acute severe colitis predict colectomy in pediatric patients.

METHODS

Patients were identified retrospectively through the gastroenterology and pathology databases. Demographic information, duration of disease, laboratory data, endoscopic appearance at scope, and histologic features of inflammation were reviewed along with medical therapies. Patients who underwent surgery within 90 days of hospitalization were compared to those who did not.

RESULTS

Fifty patients with acute severe colitis, defined as Pediatric Ulcerative Colitis Activity Index ≥65, were included. Sixteen patients had colectomies performed within 90 days of presentation. No statistically significant difference was found between the surgery and no-surgery groups for patient age, albumin, hemoglobin, or C-reactive protein, though hemoglobin trended toward significance, P = .05. The endoscopic Mayo score and histologic features of inflammation (architectural changes, chronic inflammation, eosinophils, neutrophils within the lamina propria, neutrophils in epithelium, crypt destruction, and ulceration) were similar between groups.

CONCLUSION

In pediatric patients presenting for hospitalization with acute severe colitis, no histologic features of inflammation predicted colectomy within 90 days.

HyperKvasir, a comprehensive multi-class image and video dataset for gastrointestinal endoscopy

Artificial intelligence is currently a hot topic in medicine. However, medical data is often sparse and hard to obtain due to legal restrictions and lack of medical personnel for the cumbersome and tedious process to manually label training data. These constraints make it difficult to develop systems for automatic analysis, like detecting disease or other lesions. In this respect, this article presents HyperKvasir, the largest image and video dataset of the gastrointestinal tract available today. The data is collected during real gastro- and colonoscopy examinations at Bærum Hospital in Norway and partly labeled by experienced gastrointestinal endoscopists. The dataset contains 110,079 images and 374 videos, and represents anatomical landmarks as well as pathological and normal findings. The total number of images and video frames together is around 1 million. Initial experiments demonstrate the potential benefits of artificial intelligence-based computer-assisted diagnosis systems. The HyperKvasir dataset can play a valuable role in developing better algorithms and computer-assisted examination systems not only for gastro- and colonoscopy, but also for other fields in medicine.

Application of Deep Learning for Early Screening of Colorectal Precancerous Lesions under White Light Endoscopy. Comput Math Methods Med. 2020;2020:8374317. [PMID: 32952602 PMCID: PMC7480430 DOI: 10.1155/2020/8374317]

Methods

We collected and sorted out the white light endoscopic images of some patients undergoing colonoscopy. The convolutional neural network model is used to detect whether the image contains lesions: CRC, colorectal adenoma (CRA), and colorectal polyps. The accuracy, sensitivity, and specificity rates are used as indicators to evaluate the model. Then, the instance segmentation model is used to locate and classify the lesions on the images containing lesions, and mAP (mean average precision), AP₅₀, and AP₇₅ are used to evaluate the performance of an instance segmentation model.

Results

In the process of detecting whether the image contains lesions, we compared ResNet50 with the other four models, that is, AlexNet, VGG19, ResNet18, and GoogLeNet. The result is that ResNet50 performs better than several other models. It scored an accuracy of 93.0%, a sensitivity of 94.3%, and a specificity of 90.6%. In the process of localization and classification of the lesion in images containing lesions by Mask R-CNN, its mAP, AP₅₀, and AP₇₅ were 0.676, 0.903, and 0.833, respectively.

Conclusion

We developed and compared five models for the detection of lesions in white light endoscopic images. ResNet50 showed the optimal performance, and Mask R-CNN model could be used to locate and classify lesions in images containing lesions.

Primary Humoral Immune Deficiencies: Overlooked Mimickers of Chronic Immune-Mediated Gastrointestinal Diseases in Adults

In recent years, the incidence of immune-mediated gastrointestinal disorders, including celiac disease (CeD) and inflammatory bowel disease (IBD), is increasingly growing worldwide. This generates a need to elucidate the conditions that may compromise the diagnosis and treatment of such gastrointestinal disorders. It is well established that primary immunodeficiencies (PIDs) exhibit gastrointestinal manifestations and mimic other diseases, including CeD and IBD. PIDs are often considered pediatric ailments, whereas between 25 and 45% of PIDs are diagnosed in adults. The most common PIDs in adults are the selective immunoglobulin A deficiency (SIgAD) and the common variable immunodeficiency (CVID). A trend to autoimmunity occurs, while gastrointestinal disorders are common in both diseases. Besides, the occurrence of CeD and IBD in SIgAD/CVID patients is significantly higher than in the general population. However, some differences concerning diagnostics and management between enteropathy/colitis in PIDs, as compared to idiopathic forms of CeD/IBD, have been described. There is an ongoing discussion whether CeD and IBD in CVID patients should be considered a true CeD and IBD or just CeD-like and IBD-like diseases. This review addresses the current state of the art of the most common primary immunodeficiencies in adults and co-occurring CeD and IBD.

An Extensive Study on Cross-Dataset Bias and Evaluation Metrics Interpretation for Machine Learning Applied to Gastrointestinal Tract Abnormality Classification

Precise and efficient automated identification of gastrointestinal (GI) tract diseases can help doctors treat more patients and improve the rate of disease detection and identification. Currently, automatic analysis of diseases in the GI tract is a hot topic in both computer science and medical-related journals. Nevertheless, the evaluation of such an automatic analysis is often incomplete or simply wrong. Algorithms are often only tested on small and biased datasets, and cross-dataset evaluations are rarely performed. A clear understanding of evaluation metrics and machine learning models with cross datasets is crucial to bring research in the field to a new quality level. Toward this goal, we present comprehensive evaluations of five distinct machine learning models using global features and deep neural networks that can classify 16 different key types of GI tract conditions, including pathological findings, anatomical landmarks, polyp removal conditions, and normal findings from images captured by common GI tract examination instruments. In our evaluation, we introduce performance hexagons using six performance metrics, such as recall, precision, specificity, accuracy, F1-score, and the Matthews correlation coefficient to demonstrate how to determine the real capabilities of models rather than evaluating them shallowly. Furthermore, we perform cross-dataset evaluations using different datasets for training and testing. With these cross-dataset evaluations, we demonstrate the challenge of actually building a generalizable model that could be used across different hospitals. Our experiments clearly show that more sophisticated performance metrics and evaluation methods need to be applied to get reliable models rather than depending on evaluations of the splits of the same dataset—that is, the performance metrics should always be interpreted together rather than relying on a single metric.

Inter-Observer Agreement of a New Endoscopic Score for Ulcerative Colitis Activity: Preliminary Experience

Ulcerative colitis (UC) endoscopic scores translate mucosal damage into values standardizing image analysis. Due to potential limits of current endoscopic activity indexes, we have elaborated on a new score, the “Extended Mayo Endoscopic Score (EMES),” and evaluated its inter-observer agreement in a multicenter endoscopy team, comparing concordance with the Mayo subscore. Sixteen UC consecutive patients underwent follow-up colonoscopy. Recorded videos were anonymously loaded on a web platform. Thirteen expert endoscopists evaluated UC activity using both Mayo and EMES. EMES was described in every colon segment: erythema (0: absent, 1: mild, 2: moderate, 3: severe), vascular pattern (0: normal, 1: reduction, 2: disappearance), erosions and ulcers (0: absent, 1: from 1 to 5, 2: 6 to 10, 3: >10). Weighted Fleiss’ kappa with 95% confidence interval (CI) and p-value defined inter-rater agreement. Global inter-observer agreement of EMES was moderate (kappa = 0.56, 95% CI = 0.46–0.67, p < 0.001). The evaluation of each colonic segment showed moderate agreement for all segments: ascending (kappa = 0.46, 95% CI = 0.32–0.60, p < 0.001), transverse (kappa = 0.48, 95% CI = 0.29–0.67, p < 0.001); descending (kappa = 0.49, 95% CI = 0.35–0.64, p < 0.001), sigmoid (kappa = 0.52, 95% CI = 0.39–0.65, p < 0.001) and rectum (kappa = 0.55, 95% CI = 0.42–0.69, p < 0.001). Mayo subscore agreement was similar to global EMES (kappa = 0.53, 95% CI = 0.39–0.66, p = 0.001). Therefore, our report emphasizes the importance of assessing inter-observer agreement for EMES, but also for other known scoring systems, including the Mayo subscore.

Deep learning in gastric tissue diseases: a systematic review

Background

In recent years, deep learning has gained remarkable attention in medical image analysis due to its capacity to provide results comparable to specialists and, in some cases, surpass them. Despite the emergence of deep learning research on gastric tissues diseases, few intensive reviews are addressing this topic.

Method

We performed a systematic review related to applications of deep learning in gastric tissue disease analysis by digital histology, endoscopy and radiology images.

Conclusions

This review highlighted the high potential and shortcomings in deep learning research studies applied to gastric cancer, ulcer, gastritis and non-malignant diseases. Our results demonstrate the effectiveness of gastric tissue analysis by deep learning applications. Moreover, we also identified gaps of evaluation metrics, and image collection availability, therefore, impacting experimental reproducibility.

Artificial Intelligence and Polyp Detection

PURPOSE OF REVIEW

This review highlights the history, recent advances, and ongoing challenges of artificial intelligence (AI) technology in colonic polyp detection.

RECENT FINDINGS

Hand-crafted AI algorithms have recently given way to convolutional neural networks with the ability to detect polyps in real-time. The first randomized controlled trial comparing an AI system to standard colonoscopy found a 9% increase in adenoma detection rate, but the improvement was restricted to polyps smaller than 10 mm and the results need validation. As this field rapidly evolves, important issues to consider include standardization of outcomes, dataset availability, real-world applications, and regulatory approval.

SUMMARY

AI has shown great potential for improving colonic polyp detection while requiring minimal training for endoscopists. The question of when AI will enter endoscopic practice depends on whether the technology can be integrated into existing hardware and an assessment of its added value for patient care.

Transmural impedance detects graded changes of inflammation in experimental colitis

Ulcerative colitis is a chronic disease in which the mucosa of the colon or rectum becomes inflamed. An objective biomarker of inflammation will provide quantitative measures to support qualitative assessment during an endoscopic examination. Previous studies show that transmural electrical impedance is a quantifiable biomarker of inflammation. Here, we hypothesize that impedance detects spatially restricted areas of inflammation, thereby allowing the distinction between regions that differ in their severity of inflammation. A platinum ball electrode was placed into minimally inflamed (i.e. normal) or 2,4,6-trinitrobenzene sulphonic acid (TNBS)-inflamed colonic regions of rats and impedance measurements obtained by passing current between the intraluminal and subcutaneous return electrode. Histology of the colon was correlated with impedance measurements. The impedance of minimally inflamed (normal) tissue was 1.5-1.9 kΩ. Following TNBS injection, impedance significantly decreased within the inflammatory penumbra (p < 0.05), and decreased more in the inflammatory epicentre (p = 0.02). Histological damage correlated with impedance values (p < 0.05). Thus, impedance values of 1.5-1.9, 1.3-1.4 and 0.9-1.1 kΩ corresponded to minimally inflamed, mildly inflamed and moderately inflamed tissue, respectively. In conclusion, transmural impedance is an objective, spatially localized biomarker of mucosal integrity, and distinguishes between severities of intestinal inflammation.

An overview of deep learning algorithms and water exchange in colonoscopy in improving adenoma detection

Introduction: Among the Gastrointestinal (GI) Endoscopy Editorial Board top 10 topics in advances in endoscopy in 2018, water exchange colonoscopy and artificial intelligence were both considered important advances. Artificial intelligence holds the potential to increase and water exchange significantly increases adenoma detection.Areas covered: The authors searched MEDLINE (1998-2019) using the following medical subject terms: water-aided, water-assisted and water exchange colonoscopy, adenoma, artificial intelligence, deep learning, computer-assisted detection, and neural networks. Additional related studies were manually searched from the reference lists of publications. Only fully published journal articles in English were reviewed. The latest date of the search was Aug10, 2019. Artificial intelligence, machine learning, and deep learning contribute to the promise of real-time computer-aided detection diagnosis. By emphasizing near-complete suction of infused water during insertion, water exchange provides salvage cleaning and decreases cleaning-related multi-tasking distractions during withdrawal, increasing adenoma detection. The review will address how artificial intelligence and water exchange can complement each other in improving adenoma detection during colonoscopy.Expert opinion: In 5 years, research on artificial intelligence will likely achieve real-time application and evaluation of factors contributing to quality colonoscopy. Better understanding and more widespread use of water exchange will be possible.

Towards risk-stratified colorectal cancer screening. Adding risk factors to the fecal immunochemical test: Evidence, evolution and expectations

With increasing incidence and mortality, colorectal cancer (CRC) is a growing health problem worldwide. An effective way to address CRC is by screening for fecal (occult) blood by the fecal immunochemical test (FIT). However, there is room for improvement since precursor lesions and CRC bleed intermittent and can therefore be missed by the FIT (false negatives) or, the detected blood did not result from precursor lesions or CRC (false positives). This review provides the latest evidence on risk prediction models using FIT combined with additional risk factors before colonoscopy, which risk factors to include and if these models will better discriminate between normal findings and CRC compared to the FIT-only. Many prediction models are known for CRC, but compared to the FIT, these are less effective in detecting CRC. The literature search resulted in 645 titles where 11 papers matched the inclusion criteria and were analyzed. Comparing the FIT-only with the risk prediction models for detecting CRC resulted in a significantly increased discrimination for the models. In addition, 2 different risk-stratification categories before colonoscopy were distinguished, namely the 1-model approach which combined risk factors with FIT results in a prediction model while the 2 step approach used risk factors apart from the FIT. Finally, combining FIT with CRC risk factors by means of a model before colonoscopy seems effective regarding discriminative power, however, more research is needed for validation combined with transparent and standardized reporting to improve quality assessment, for which suggestions are reported in this study.

Physically active individuals have a 23% lower risk of any colorectal neoplasia and a 27% lower risk of advanced colorectal neoplasia than their non-active counterparts: systematic review and meta-analysis of observational studies

BACKGROUND

Few studies have examined the associations between physical activity (PA), sedentary behaviour (SB) and risk of colorectal neoplasia (CN).

METHODS

We systematically searched Medline, Embase, PsyInfo, Cochrane and other sources from their inception to 30 September 2018 for cohort, case-control and cross-sectional studies that evaluated these associations in asymptomatic, average-risk subjects. Random-effect models were used to estimate relative risks (RRs) of any-type CN, advanced CN, and non-advanced CN, respectively, in individuals with the highest versus the lowest level of PA and SB. Dose-response analyses and subgroup analyses were conducted. The I² statistic was used to examine heterogeneity among studies.

RESULTS

We identified 32 observational studies, including 17 cross-sectional studies, 10 case-control studies and five longitudinal studies. PA (highest vs lowest) was inversely associated with risk for any-type CN (n=23 studies) and advanced CN (n=15 studies), with a RR of 0.77 (95% CI=0.71 to 0.83, I²=57.5%) and 0.73 (95% CI=0.63 to 0.82, I²=45.5%), respectively. There was no association between PA and non-advanced CN (n=5 studies). There was an as association between PA and any-type CN in both sexes, and also for the distal colon. We found no dose-response relationship between PA and any-type or advanced CN. Based on three studies identified, SB time (longest vs shortest) was associated with an increased risk of advanced CN (RR=1.24, 95% CI 1.04 to 1.49, I²=14.4%). No publication bias was detected by Begg's test.

CONCLUSION

We report a 23% lower relative risk of any type of CN and a 27% lower risk of advanced CN in people with the highest level of PA compared with those in the lowest.

Performance of a Deep Learning Model vs Human Reviewers in Grading Endoscopic Disease Severity of Patients With Ulcerative Colitis

IMPORTANCE

Assessing endoscopic disease severity in ulcerative colitis (UC) is a key element in determining therapeutic response, but its use in clinical practice is limited by the requirement for experienced human reviewers.

OBJECTIVE

To determine whether deep learning models can grade the endoscopic severity of UC as well as experienced human reviewers.

DESIGN, SETTING, AND PARTICIPANTS

In this diagnostic study, retrospective grading of endoscopic images using the 4-level Mayo subscore was performed by 2 independent reviewers with score discrepancies adjudicated by a third reviewer. Using 16 514 images from 3082 patients with UC who underwent colonoscopy at a single tertiary care referral center in the United States between January 1, 2007, and December 31, 2017, a 159-layer convolutional neural network (CNN) was constructed as a deep learning model to train and categorize images into 2 clinically relevant groups: remission (Mayo subscore 0 or 1) and moderate to severe disease (Mayo subscore, 2 or 3). Ninety percent of the cohort was used to build the model and 10% was used to test it; the process was repeated 10 times. A set of 30 full-motion colonoscopy videos, unseen by the model, was then used for external validation to mimic real-world application.

MAIN OUTCOMES AND MEASURES

Model performance was assessed using area under the receiver operating curve (AUROC), sensitivity and specificity, positive predictive value (PPV), and negative predictive value (NPV). Kappa statistics (κ) were used to measure agreement of the CNN relative to adjudicated human reference cores.

RESULTS

The authors included 16 514 images from 3082 unique patients (median [IQR] age, 41.3 [26.1-61.8] years, 1678 [54.4%] female), with 3980 images (24.1%) classified as moderate-to-severe disease by the adjudicated reference score. The CNN was excellent for distinguishing endoscopic remission from moderate-to-severe disease with an AUROC of 0.966 (95% CI, 0.967-0.972); a PPV of 0.87 (95% CI, 0.85-0.88) with a sensitivity of 83.0% (95% CI, 80.8%-85.4%) and specificty of 96.0% (95% CI, 95.1%-97.1%); and NPV of 0.94 (95% CI, 0.93-0.95). Weighted κ agreement between the CNN and the adjudicated reference score was also good for identifying exact Mayo subscores (κ = 0.84; 95% CI, 0.83-0.86) and was similar to the agreement between experienced reviewers (κ = 0.86; 95% CI, 0.85-0.87). Applying the CNN to entire colonoscopy videos had similar accuracy for identifying moderate to severe disease (AUROC, 0.97; 95% CI, 0.963-0.969).

CONCLUSIONS AND RELEVANCE

This study found that deep learning model performance was similar to experienced human reviewers in grading endoscopic severity of UC. Given its scalability, this approach could improve the use of colonoscopy for UC in both research and routine practice.

A comprehensive review and update on ulcerative colitis. Dis Mon. 2019;65:100851. [PMID: 30837080 DOI: 10.1016/j.disamonth.2019.02.004]

Ulcerative colitis (UC) is a chronic idiopathic inflammatory bowel disorder of the colon that causes continuous mucosal inflammation extending from the rectum to the more proximal colon, with variable extents. UC is characterized by a relapsing and remitting course. UC was first described by Samuel Wilks in 1859 and it is more common than Crohn's disease worldwide. The overall incidence and prevalence of UC is reported to be 1.2-20.3 and 7.6-245 cases per 100,000 persons/year respectively. UC has a bimodal age distribution with an incidence peak in the 2nd or 3rd decades and followed by second peak between 50 and 80 years of age. The key risk factors for UC include genetics, environmental factors, autoimmunity and gut microbiota. The classic presentation of UC include bloody diarrhea with or without mucus, rectal urgency, tenesmus, and variable degrees of abdominal pain that is often relieved by defecation. UC is diagnosed based on the combination of clinical presentation, endoscopic findings, histology, and the absence of alternative diagnoses. In addition to confirming the diagnosis of UC, it is also important to define the extent and severity of inflammation, which aids in the selection of appropriate treatment and for predicting the patient's prognosis. Ileocolonoscopy with biopsy is the only way to make a definitive diagnosis of UC. A pathognomonic finding of UC is the presence of continuous colonic inflammation characterized by erythema, loss of normal vascular pattern, granularity, erosions, friability, bleeding, and ulcerations, with distinct demarcation between inflamed and non-inflamed bowel. Histopathology is the definitive tool in diagnosing UC, assessing the disease severity and identifying intraepithelial neoplasia (dysplasia) or cancer. The classical histological changes in UC include decreased crypt density, crypt architectural distortion, irregular mucosal surface and heavy diffuse transmucosal inflammation, in the absence of genuine granulomas. Abdominal computed tomographic (CT) scanning is the preferred initial radiographic imaging study in UC patients with acute abdominal symptoms. The hallmark CT finding of UC is mural thickening with a mean wall thickness of 8 mm, as opposed to a 2-3 mm mean wall thickness of the normal colon. The Mayo scoring system is a commonly used index to assess disease severity and monitor patients during therapy. The goals of treatment in UC are three fold-improve quality of life, achieve steroid free remission and minimize the risk of cancer. The choice of treatment depends on disease extent, severity and the course of the disease. For proctitis, topical 5-aminosalicylic acid (5-ASA) drugs are used as the first line agents. UC patients with more extensive or severe disease should be treated with a combination of oral and topical 5-ASA drugs +/- corticosteroids to induce remission. Patients with severe UC need to be hospitalized for treatment. The options in these patients include intravenous steroids and if refractory, calcineurin inhibitors (cyclosporine, tacrolimus) or tumor necrosis factor-α antibodies (infliximab) are utilized. Once remission is induced, patients are then continued on appropriate medications to maintain remission. Indications for emergency surgery include refractory toxic megacolon, colonic perforation, or severe colorectal bleeding.

Strong associations of a healthy lifestyle with all stages of colorectal carcinogenesis: Results from a large cohort of participants of screening colonoscopy

The risk of developing colorectal cancer (CRC) is associated with a wide range of dietary and lifestyle factors. The individual contribution of single modifiable factors, such as alcohol consumption, physical activity, smoking, body mass index (BMI) or dietary components, to the development of CRC has been investigated extensively, but evidence on their combined effect at various stages of colorectal carcinogenesis is sparse. The aim of our study was to analyze the association of a healthy lifestyle pattern with prevalence of early and advanced colorectal neoplasms. A total of 13,600 participants of screening colonoscopy in Saarland/Germany (mean age 62.9 years) who were enrolled in the KolosSal study (Effektivität der Früherkennungs-Koloskopie: eine Saarland-weite Studie) from 2005 until 2013 were included in this cross-sectional analysis. Dietary and lifestyle data were collected and colonoscopy results were extracted from physicians' reports. The association of an a priori defined healthy lifestyle score-including dietary intake, alcohol consumption, physical activity, smoking and BMI-with early and advanced colorectal neoplasms was assessed by multiple logistic regression analyses with comprehensive adjustment for potential confounders. Strong inverse dose-response relationships were observed between an overall healthier lifestyle pattern and presence of advanced colorectal neoplasms, nonadvanced adenomas and hyperplastic polyps (p value <0.0001 in all cases), with adjusted odds ratios (95% CI) for the highest compared to the lowest category of the healthy lifestyle score of 0.41 (0.30-0.56), 0.42 (0.33-0.54) and 0.39 (0.29-0.54) respectively. A healthy lifestyle is strongly associated with lower risk of all stages of colorectal neoplasms.

Improving Cancer Preventive Behaviors: A Randomized Trial of Tailored Lifestyle Feedback in Colorectal Cancer Screening

BACKGROUND

Cancer screening provides an opportunity to increase awareness of cancer-preventive lifestyle behaviors such as nonsmoking, physical activity, low alcohol consumption, and a healthy diet. We tested the effect of standardized, individually tailored written feedback (TF), and a standard leaflet (SL) on 1-year lifestyle behaviors in a colorectal cancer screening setting.

METHODS

A total of 3,642 men and women aged 50-74 years invited to sigmoidoscopy screening were randomly assigned to: (i) TF; (ii) SL for cancer-preventive lifestyle behaviors; or (iii) control. Participants were mailed two self-reported lifestyle questionnaires (LSQ) 1 year apart. The TF intervention was based on the prescreening LSQ answers. We analyzed differences [with 95% confidence intervals (CI)] by comparing prescreening to 1-year follow-up of single cancer-preventive factors and the number of cancer-preventive lifestyle behaviors (range 0-4) between the groups by multivariable logistic regression and analysis of covariance (ANCOVA).

RESULTS

A total of 1,054 screening participants without neoplastic findings (29% of those invited to screening) were included in this study. Participants in the TF group increased their number of cancer-preventive lifestyle behaviors significantly compared with those in the control group by 0.11 (95% CI, 0.02 to 0.19). Overweight/obese individuals in the TF group had a -0.84 kg (95% CI, -1.47 to -0.22) larger reduction in body weight compared with the control group.

CONCLUSIONS

TF at sigmoidoscopy screening led to small improvements in cancer-preventive behaviors.

IMPACT

Colorectal cancer screening is a suitable setting for increasing awareness of cancer-preventive behavior.