Artificial Intelligence Applied to Colonoscopy: Is It Time to Take a Step Forward?

Development, deployment and scaling of operating room-ready artificial intelligence for real-time surgical decision support

Deep learning for computer vision can be leveraged for interpreting surgical scenes and providing surgeons with real-time guidance to avoid complications. However, neither generalizability nor scalability of computer-vision-based surgical guidance systems have been demonstrated, especially to geographic locations that lack hardware and infrastructure necessary for real-time inference. We propose a new equipment-agnostic framework for real-time use in operating suites. Using laparoscopic cholecystectomy and semantic segmentation models for predicting safe/dangerous ("Go"/"No-Go") zones of dissection as an example use case, this study aimed to develop and test the performance of a novel data pipeline linked to a web-platform that enables real-time deployment from any edge device. To test this infrastructure and demonstrate its scalability and generalizability, lightweight U-Net and SegFormer models were trained on annotated frames from a large and diverse multicenter dataset from 136 institutions, and then tested on a separate prospectively collected dataset. A web-platform was created to enable real-time inference on any surgical video stream, and performance was tested on and optimized for a range of network speeds. The U-Net and SegFormer models respectively achieved mean Dice scores of 57% and 60%, precision 45% and 53%, and recall 82% and 75% for predicting the Go zone, and mean Dice scores of 76% and 76%, precision 68% and 68%, and recall 92% and 92% for predicting the No-Go zone. After optimization of the client-server interaction over the network, we deliver a prediction stream of at least 60 fps and with a maximum round-trip delay of 70 ms for speeds above 8 Mbps. Clinical deployment of machine learning models for surgical guidance is feasible and cost-effective using a generalizable, scalable and equipment-agnostic framework that lacks dependency on hardware with high computing performance or ultra-fast internet connection speed.

Diagnostic Accuracy of Artificial Intelligence in Endoscopy: Umbrella Review

Background

Some research has already reported the diagnostic value of artificial intelligence (AI) in different endoscopy outcomes. However, the evidence is confusing and of varying quality.

Objective

This review aimed to comprehensively evaluate the credibility of the evidence of AI's diagnostic accuracy in endoscopy.

Methods

Before the study began, the protocol was registered on PROSPERO (CRD42023483073). First, 2 researchers searched PubMed, Web of Science, Embase, and Cochrane Library using comprehensive search terms. Then, researchers screened the articles and extracted information. We used A Measurement Tool to Assess Systematic Reviews 2 (AMSTAR2) to evaluate the quality of the articles. When there were multiple studies aiming at the same result, we chose the study with higher-quality evaluations for further analysis. To ensure the reliability of the conclusions, we recalculated each outcome. Finally, the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) was used to evaluate the credibility of the outcomes.

Results

A total of 21 studies were included for analysis. Through AMSTAR2, it was found that 8 research methodologies were of moderate quality, while other studies were regarded as having low or critically low quality. The sensitivity and specificity of 17 different outcomes were analyzed. There were 4 studies on esophagus, 4 studies on stomach, and 4 studies on colorectal regions. Two studies were associated with capsule endoscopy, two were related to laryngoscopy, and one was related to ultrasonic endoscopy. In terms of sensitivity, gastroesophageal reflux disease had the highest accuracy rate, reaching 97%, while the invasion depth of colon neoplasia, with 71%, had the lowest accuracy rate. On the other hand, the specificity of colorectal cancer was the highest, reaching 98%, while the gastrointestinal stromal tumor, with only 80%, had the lowest specificity. The GRADE evaluation suggested that the reliability of most outcomes was low or very low.

Conclusions

AI proved valuabe in endoscopic diagnoses, especially in esophageal and colorectal diseases. These findings provide a theoretical basis for developing and evaluating AI-assisted systems, which are aimed at assisting endoscopists in carrying out examinations, leading to improved patient health outcomes. However, further high-quality research is needed in the future to fully validate AI's effectiveness.

GI genius endoscopy module: a clinical profile

INTRODUCTION

The identification of early-stage colorectal cancers (CRC) and the resection of pre-cancerous neoplastic lesions through colonoscopy allows to decrease both CRC incidence and mortality. However, colonoscopy miss rates up to 26% for adenomas and 9% for advanced adenomas have been reported. In recent years, artificial intelligence (AI) systems have been emerging as easy-to-use tools, potentially lowering the risk of missing lesions.

AREAS COVERED

This review paper focuses on GI Genius device (Medtronic Co. Minneapolis, MN, U.S.A.) a computer-assisted tool designed to assist endoscopists during standard white-light colonoscopies in detecting mucosal lesions.

EXPERT OPINION

Randomized controlled trials (RCTs) suggest that GI Genius is a safe and effective tool for improving adenoma detection, especially in CRC screening and surveillance colonoscopies. However, its impact seems to be less significant among experienced endoscopists and in real-world clinical scenarios compared to the controlled conditions of RCTs. Furthermore, it appears that GI Genius mainly enhances the detection of non-advanced, small polyps, but does not significantly impact the identification of advanced and difficult-to-detect adenoma. When using GI Genius, no complications were documented. Only a small number of studies reported an increased in withdrawal time or the removal of non-neoplastic lesions.

Pathology and Clinical Significance of Inflammatory Bowel Disease-Associated Colorectal Dysplastic Lesions

Timely diagnosis and effective management of colorectal dysplasia play a vital role in preventing mortality from colorectal cancer in patients with chronic inflammatory bowel disease. This review provides a contemporary overview of the pathologic and endoscopic classification of dysplasia in inflammatory bowel disease, their roles in determining surveillance and management algorithms, and emerging diagnostic and therapeutic approaches that might further enhance patient management.

Comparative Study of Predictive Models for the Detection of Patients at High Risk of Inadequate Colonic Cleansing

Background: Various predictive models have been published to identify outpatients with inadequate colonic cleansing who may benefit from intensified preparations to improve colonoscopy quality. The main objective of this study was to compare the accuracy of three predictive models for identifying poor bowel preparation in outpatients undergoing colonoscopy. Methods: This cross-sectional study included patients scheduled for outpatient colonoscopy over a 3-month period. We evaluated and compared three predictive models (Models 1-3). The quality of colonic cleansing was assessed using the Boston Bowel Preparation Scale. We calculated the area under the curve (AUC) and the corresponding 95% confidence interval for each model. Additionally, we performed simple and multiple logistic regression analyses to identify variables associated with inadequate colonic cleansing and developed a new model. Results: A total of 649 consecutive patients were included in the study, of whom 84.3% had adequate colonic cleansing quality. The AUCs of Model 1 (AUC = 0.67, 95% CI [0.63-0.70]) and Model 2 (AUC = 0.62, 95% CI [0.58-0.66]) were significantly higher than that of Model 3 (AUC = 0.54, 95% CI [0.50-0.58]; p < 0.001). Moreover, Model 1 outperformed Model 2 (p = 0.013). However, the new model did not demonstrate improved accuracy compared to the older models (AUC = 0.671). Conclusions: Among the three compared models, Model 1 showed the highest accuracy for predicting poor bowel preparation in outpatients undergoing colonoscopy and could be useful in clinical practice to decrease the percentage of inadequately prepared patients.

Colon Bowel Preparation in the Era of Artificial Intelligence: Is There Potential for Enhancing Colon Bowel Cleansing?

BACKGROUND AND OBJECTIVES

Proper bowel preparation is of paramount importance for enhancing adenoma detection rates and reducing postcolonoscopic colorectal cancer risk. Despite recommendations from gastroenterology societies regarding the optimal rates of successful bowel preparation, these guidelines are frequently unmet. Various approaches have been employed to enhance the rates of successful bowel preparation, yet the quality of cleansing remains suboptimal. Intensive bowel preparation techniques, supplementary administration of bowel solutions, and educational interventions aimed at improving patient adherence to instructions have been commonly utilized, particularly among patients at a high risk of inadequate bowel preparation. Expedited strategies conducted on the same day as the procedure have also been endorsed by scientific organizations. More recently, the utilization of artificial intelligence (AI) has emerged for the preprocedural detection of inadequate bowel preparation, holding the potential to guide the preparation process immediately preceding colonoscopy. This manuscript comprehensively reviews the current strategies employed to optimize bowel cleansing, with a specific focus on patients with elevated risks for inadequate bowel preparation. Additionally, the prospective role of AI in this context is thoroughly examined.

CONCLUSIONS

While a majority of outpatients may achieve cleanliness with standard cleansing protocols, dealing with hard-to-prepare patients remains a challenge. Rescue strategies based on AI are promising, but such evidence remains limited. To ensure proper bowel cleansing, a combination of strategies should be performed.

Artificial Intelligence-Assisted Optical Diagnosis: A Comprehensive Review of Its Role in Leave-In-Situ and Resect-and-Discard Strategies in Colonoscopy

Colorectal cancer screening plays a vital role in early detection and removal of precancerous adenomas, contributing to decreased mortality rates. Most polyps found during colonoscopies are small and unlikely to harbor advanced neoplasia or invasive cancer, leading to the development of "leave-in-situ" and "resect-and-discard" approaches. These strategies could lead to significant cost savings and efficiencies, but their implementation has been hampered by concerns around financial incentives, medical-legal risks, and local rules for tissue handling. This article reviews the potential of artificial intelligence to enhance the accuracy of polyp diagnosis through computer-aided diagnosis (CADx). While the adoption of CADx in optical biopsy has shown mixed results, it has the potential to significantly improve the management of colorectal polyps. Several studies reviewed in this article highlight the varied results of CADx in optical biopsy for colorectal polyps. Although artificial intelligence does not consistently outperform expert endoscopists, it has the potential to serve as a beneficial secondary reader, aiding in accurate optical diagnosis and increasing the confidence of the endoscopist. These studies indicate that although CADx holds great potential, it is yet to fully meet the performance thresholds necessary for clinical implementation.

Computer-Based Diagnosis of Celiac Disease by Quantitative Processing of Duodenal Endoscopy Images

Celiac disease (CD) is a lifelong chronic autoimmune systemic disease that primarily affects the small bowel of genetically susceptible individuals. The diagnostics of adult CD currently rely on specific serology and the histological assessment of duodenal mucosa on samples taken by upper digestive endoscopy. Because of several pitfalls associated with duodenal biopsy sampling and histopathology, and considering the pediatric no-biopsy diagnostic criteria, a biopsy-avoiding strategy has been proposed for adult CD diagnosis also. Several endoscopic changes have been reported in the duodenum of CD patients, as markers of villous atrophy (VA), with good correlation with serology. In this setting, an opportunity lies in the automated detection of these endoscopic markers, during routine endoscopy examinations, as potential case-finding of unsuspected CD. We collected duodenal endoscopy images from 18 CD newly diagnosed CD patients and 16 non-CD controls and applied machine learning (ML) and deep learning (DL) algorithms on image patches for the detection of VA. Using histology as standard, high diagnostic accuracy was seen for all algorithms tested, with the layered convolutional neural network (CNN) having the best performance, with 99.67% sensitivity and 98.07% positive predictive value. In this pilot study, we provide an accurate algorithm for automated detection of mucosal changes associated with VA in CD patients, compared to normally appearing non-atrophic mucosa in non-CD controls, using histology as a reference.

Computer-Based Diagnosis of Celiac Disease by Quantitative Processing of Duodenal Endoscopy Images

Celiac disease (CD) is a lifelong chronic autoimmune systemic disease that primarily affects the small bowel of genetically susceptible individuals. The diagnostics of adult CD currently rely on specific serology and the histological assessment of duodenal mucosa on samples taken by upper digestive endoscopy. Because of several pitfalls associated with duodenal biopsy sampling and histopathology, and considering the pediatric no-biopsy diagnostic criteria, a biopsy-avoiding strategy has been proposed for adult CD diagnosis also. Several endoscopic changes have been reported in the duodenum of CD patients, as markers of villous atrophy (VA), with good correlation with serology. In this setting, an opportunity lies in the automated detection of these endoscopic markers, during routine endoscopy examinations, as potential case-finding of unsuspected CD. We collected duodenal endoscopy images from 18 CD newly diagnosed CD patients and 16 non-CD controls and applied machine learning (ML) and deep learning (DL) algorithms on image patches for the detection of VA. Using histology as standard, high diagnostic accuracy was seen for all algorithms tested, with the layered convolutional neural network (CNN) having the best performance, with 99.67% sensitivity and 98.07% positive predictive value. In this pilot study, we provide an accurate algorithm for automated detection of mucosal changes associated with VA in CD patients, compared to normally appearing non-atrophic mucosa in non-CD controls, using histology as a reference.