How should we do colon capsule endoscopy reading: a practical guide

Anatomical landmarks localization for capsule endoscopy studies

Wireless Capsule Endoscopy is a medical procedure that uses a small, wireless camera to capture images of the inside of the digestive tract. The identification of the entrance and exit of the small bowel and of the large intestine is one of the first tasks that need to be accomplished to read a video. This paper addresses the design of a clinical decision support tool to detect these anatomical landmarks. We have developed a system based on deep learning that combines images, timestamps, and motion data to achieve state-of-the-art results. Our method does not only classify the images as being inside or outside the studied organs, but it is also able to identify the entrance and exit frames. The experiments performed with three different datasets (one public and two private) show that our system is able to approximate the landmarks while achieving high accuracy on the classification problem (inside/outside of the organ). When comparing the entrance and exit of the studied organs, the distance between predicted and real landmarks is reduced from 1.5 to 10 times with respect to previous state-of-the-art methods.

The intersection of video capsule endoscopy and artificial intelligence: addressing unique challenges using machine learning

Introduction

Technical burdens and time-intensive review processes limit the practical utility of video capsule endoscopy (VCE). Artificial intelligence (AI) is poised to address these limitations, but the intersection of AI and VCE reveals challenges that must first be overcome. We identified five challenges to address. Challenge #1: VCE data are stochastic and contains significant artifact. Challenge #2: VCE interpretation is cost-intensive. Challenge #3: VCE data are inherently imbalanced. Challenge #4: Existing VCE AIMLT are computationally cumbersome. Challenge #5: Clinicians are hesitant to accept AIMLT that cannot explain their process.

Methods

An anatomic landmark detection model was used to test the application of convolutional neural networks (CNNs) to the task of classifying VCE data. We also created a tool that assists in expert annotation of VCE data. We then created more elaborate models using different approaches including a multi-frame approach, a CNN based on graph representation, and a few-shot approach based on meta-learning.

Results

When used on full-length VCE footage, CNNs accurately identified anatomic landmarks (99.1%), with gradient weighted-class activation mapping showing the parts of each frame that the CNN used to make its decision. The graph CNN with weakly supervised learning (accuracy 89.9%, sensitivity of 91.1%), the few-shot model (accuracy 90.8%, precision 91.4%, sensitivity 90.9%), and the multi-frame model (accuracy 97.5%, precision 91.5%, sensitivity 94.8%) performed well.

Discussion

Each of these five challenges is addressed, in part, by one of our AI-based models. Our goal of producing high performance using lightweight models that aim to improve clinician confidence was achieved.

Colon capsule endoscopy following incomplete colonoscopy in routine clinical settings

BACKGROUND

Colon capsule endoscopy (CCE) was introduced in our department on two indications; following incomplete colonoscopy as an alternative to CT colonography, and in patients with a history of incomplete colonoscopy as an alternative to anesthesia-assisted (AA) colonoscopy. We aimed to compare the quality of CCE, defined by completion rate and polyp detection rate (PDR), with that of CT colonography and AA colonoscopy, respectively.

METHODS

Patients referred for CCE from May 2020 until November 2021 were consecutively included in this prospective cohort study. Demographics, indication and CCE outcomes were registered from the electronic patient record. Completion rate and PDR in CCE as an alternative to CT colonography were compared with those of a historical cohort undergoing CT colonography following incomplete colonoscopy. Completion rate and PDR in CCE as an alternative to AA colonoscopy were compared with those of a time true parallel cohort undergoing AA colonoscopy.

RESULTS

In 65 patients undergoing CCE, 36 (57%) were referred as an alternative to CT colonography. The completion rate in this group was 44% compared to 96% in CT colonography (p < 0.001). The PDR in complete CCE in this group was 75% in CCE compared to 20% in CT colonography (p < 0.001). The remaining 27 (43%) of the sample were referred for CCE as an alternative to AA colonoscopy. The completion rate in this group was 33% compared to 100% in AA colonoscopy (p < 0.001). The PDR in complete CCE in this group was 78% in CCE compared to 35% in AA colonoscopy (p = 0.013).

CONCLUSIONS

The completion rate of CCE following incomplete colonoscopy is inferior to that of CT colonography and AA colonoscopy. The PDR of CCE was high, indicating an acceptable sensitivity in complete investigations, but in our settings the completion rate of CCE on this indication is unacceptably low.

CLINICAL TRIAL REGISTRATION

NCT04307901 (ClinicalTrials.gov, March 13, 2020).

The optimal use of colon capsule endoscopes in clinical practice

Colon capsule endoscopy (CCE) has been available for nearly two decades but has grappled with being an equal diagnostic alternative to optical colonoscopy (OC). Due to the COVID-19 pandemic, CCE has gained more foothold in clinical practice. In this cutting-edge review, we aim to present the existing knowledge on the pros and cons of CCE and discuss whether the modality is ready for a larger roll-out in clinical settings. We have included clinical trials and reviews with the most significant impact on the current position of CCE in clinical practice and discuss the challenges that persist and how they could be addressed to make CCE a more sustainable imaging modality with an adenoma detection rate equal to OC and a low re-investigation rate by a proper preselection of suitable populations. CCE is embedded with a very low risk of severe complications and can be performed in the patient's home as a pain-free procedure. The diagnostic accuracy is found to be equal to OC. However, a significant drawback is low completion rates eliciting a high re-investigation rate. Furthermore, the bowel preparation before CCE is extensive due to the high demand for clean mucosa. CCE is currently not suitable for large-scale implementation in clinical practice mainly due to high re-investigation rates. By a better preselection before CCE and the implantation of artificial intelligence for picture and video analysis, CCE could be the alternative to OC needed to move away from in-hospital services and relieve long-waiting lists for OC.

Artificial intelligence to improve polyp detection and screening time in colon capsule endoscopy

Colon Capsule Endoscopy (CCE) is a minimally invasive procedure which is increasingly being used as an alternative to conventional colonoscopy. Videos recorded by the capsule cameras are long and require one or more experts' time to review and identify polyps or other potential intestinal problems that can lead to major health issues. We developed and tested a multi-platform web application, AI-Tool, which embeds a Convolution Neural Network (CNN) to help CCE reviewers. With the help of artificial intelligence, AI-Tool is able to detect images with high probability of containing a polyp and prioritize them during the reviewing process. With the collaboration of 3 experts that reviewed 18 videos, we compared the classical linear review method using RAPID Reader Software v9.0 and the new software we present. Applying the new strategy, reviewing time was reduced by a factor of 6 and polyp detection sensitivity was increased from 81.08 to 87.80%.

Inter/Intra-Observer Agreement in Video-Capsule Endoscopy: Are We Getting It All Wrong? A Systematic Review and Meta-Analysis

Video-capsule endoscopy (VCE) reading is a time- and energy-consuming task. Agreement on findings between readers (either different or the same) is a crucial point for increasing performance and providing valid reports. The aim of this systematic review with meta-analysis is to provide an evaluation of inter/intra-observer agreement in VCE reading. A systematic literature search in PubMed, Embase and Web of Science was performed throughout September 2022. The degree of observer agreement, expressed with different test statistics, was extracted. As different statistics are not directly comparable, our analyses were stratified by type of test statistics, dividing them in groups of "None/Poor/Minimal", "Moderate/Weak/Fair", "Good/Excellent/Strong" and "Perfect/Almost perfect" to report the proportions of each. In total, 60 studies were included in the analysis, with a total of 579 comparisons. The quality of included studies, assessed with the MINORS score, was sufficient in 52/60 studies. The most common test statistics were the Kappa statistics for categorical outcomes (424 comparisons) and the intra-class correlation coefficient (ICC) for continuous outcomes (73 comparisons). In the overall comparison of inter-observer agreement, only 23% were evaluated as "good" or "perfect"; for intra-observer agreement, this was the case in 36%. Sources of heterogeneity (high, I2 81.8-98.1%) were investigated with meta-regressions, showing a possible role of country, capsule type and year of publication in Kappa inter-observer agreement. VCE reading suffers from substantial heterogeneity and sub-optimal agreement in both inter- and intra-observer evaluation. Artificial-intelligence-based tools and the adoption of a unified terminology may progressively enhance levels of agreement in VCE reading.

Colon Capsule Endoscopy in the Diagnosis of Colon Polyps: Who Needs a Colonoscopy?

Colon screening programs have reduced colon cancer mortality. Population screening should be minimally invasive, safe, acceptably sensitive, cost-effective, and scalable. The range of screening modalities include guaiac or immunochemical fecal occult blood testing and CT colonography and colonoscopy. A number of carefully controlled studies concur that second-generation capsule endoscopy has excellent sensitivity for polyp detection and a high negative predictive value. Colon capsules fulfill the screening expectation of safety, high sensitivity for polyp detection, and patient acceptance, and appear to straddle the divide between occult blood testing and colonoscopy. While meeting these criteria, there remains the challenges of scaling, capsule practitioner training, resource allocation, and implementing change of practice. Like CT colonography, capsule screening presents the clinician with a decision on the threshold for colonoscopy referral. Overall, colon capsules are an invaluable tool in polyp detection and colon screening and offer a filter that determines “who needs a colonoscopy?”.

Indications, Detection, Completion and Retention Rates of Capsule Endoscopy in Two Decades of Use: A Systematic Review and Meta-Analysis

Background: Capsule endoscopy (CE) has become a widespread modality for non-invasive evaluation of the gastrointestinal (GI) tract, with several CE models having been developed throughout the years. The aim of this systematic review and meta-analysis is to evaluate performance measures such as completion, detection and retention rates of CE. Methods: Literature through to August 2021 was screened for articles regarding all capsule types: small bowel, double-headed capsule for the colon or PillCam^®Crohn’s capsule, magnetically-controlled capsule endoscopy, esophageal capsule and patency capsule. Primary outcomes included detection rate (DR), completion rate (CR) and capsule retention rate (RR). DR, CR and RR were also analyzed in relation to indications such as obscure GI bleeding (OGIB), known/suspected Crohn’s disease (CD), celiac disease (CeD), neoplastic lesions (NL) and clinical symptoms (CS). Results: 328 original articles involving 86,930 patients who underwent CE were included. OGIB was the most common indication (n = 44,750), followed by CS (n = 17,897), CD (n = 11,299), NL (n = 4989) and CeD (n = 947). The most used capsule type was small bowel CE in 236 studies. DR, CR and RR for all indications were 59%, 89.6% and 2%, respectively. According to specific indications: DR were 55%, 66%, 63%, 52% and 62%; CR were 90.6%, 86.5%, 78.2%, 94% and 92.8%; and RR were 2%, 4%, 1%, 6% and 2%. Conclusions: Pooled DR, CR and RR are acceptable for all capsule types. OGIB is the most common indication for CE. Technological advancements have expanded the scope of CE devices in detecting GI pathology with acceptable rates for a complete examination.

How a managed service for colon capsule endoscopy works in an overstretched healthcare system

Lower gastrointestinal diagnostics have been facing significant capacity constraints, which the COVID-19 pandemic has exacerbated due to significant reductions in endoscopy procedures. Colon Capsule Endoscopy (CCE) provides a safe, viable solution to offset ongoing demand and could be a valuable tool for the recovery of endoscopy services post-COVID. NHS Scotland has already begun a country-wide rollout of CCE as a managed service, and NHS England have committed to a pilot scheme of 11,000 capsules via hospital-based delivery. Here, we outline a proven method of CCE delivery that ensures the CCE and results are delivered in an efficient, clinically robust manner with high patient acceptability levels through a managed service. Delivering CCE without a managed service is likely to be slower, more costly, and less effective, limiting the many benefits of CCE as an addition to the standard diagnostic pathway for bowel cancer.

Applicability of colon capsule endoscopy as pan-endoscopy: From bowel preparation, transit, and rating times to completion rate and patient acceptance

Background and study aims  Colon capsule endoscopy (CCE) has the potential to explore the entire gastrointestinal tract. The aim of this study was to assess the applicability of CCE as pan-endoscopy. Patients and methods  Healthy participants received CCE with bowel preparation (bisacodyl, polyethylene electrolyte glycol (PEG) + ascorbic acid) and booster regimen (metoclopramide, oral sulfate solution (OSS)). For each segment of the gastrointestinal tract, the following quality parameters were assessed: cleanliness, transit times, reading times, patient acceptance and safety of the procedure. When all gastrointestinal segments had cleansing score good or excellent, cleanliness of the whole gastrointestinal tract was assessed as good. Participants' expected and perceived burden was assessed by questionnaires and participants were asked to grade the procedure (scale 0-10). All serious adverse events (SAEs) were documented. Results  A total of 451 CCE procedures were analyzed. A good cleansing score was achieved in the stomach in 69.6%, in the SB in 99.1 % and in the colon in 76.6 %. Cleanliness of the whole gastrointestinal tract was good in 52.8 % of the participants. CCE median transit time of the whole gastrointestinal tract was 583 minutes IQR 303-659). The capsule reached the descending colon in 94.7 %. Median reading time per procedure was 70 minutes (IQR 57-83). Participants graded the procedure with a 7.8. There were no procedure-related SAEs. Conclusions  CCE as pan-endoscopy has shown to be a safe procedure with good patient acceptance. When cleanliness of all gastrointestinal segments per patient, completion rate and reading time will be improved, CCE can be applied as a good non-invasive alternative to evaluate the gastrointestinal tract.

Artificial intelligence for the detection of polyps or cancer with colon capsule endoscopy

Colorectal cancer is common and can be devastating, with long-term survival rates vastly improved by early diagnosis. Colon capsule endoscopy (CCE) is increasingly recognised as a reliable option for colonic surveillance, but widespread adoption has been slow for several reasons, including the time-consuming reading process of the CCE recording. Automated image recognition and artificial intelligence (AI) are appealing solutions in CCE. Through a review of the currently available and developmental technologies, we discuss how AI is poised to deliver at the forefront of CCE in the coming years. Current practice for CCE reporting often involves a two-step approach, with a ‘pre-reader’ and ‘validator’. This requires skilled and experienced readers with a significant time commitment. Therefore, CCE is well-positioned to reap the benefits of the ongoing digital innovation. This is likely to initially involve an automated AI check of finished CCE evaluations as a quality control measure. Once felt reliable, AI could be used in conjunction with a ‘pre-reader’, before adopting more of this role by sending provisional results and abnormal frames to the validator. With time, AI would be able to evaluate the findings more thoroughly and reduce the input required from human readers and ultimately autogenerate a highly accurate report and recommendation of therapy, if required, for any pathology identified. As with many medical fields reliant on image recognition, AI will be a welcome aid in CCE. Initially, this will be as an adjunct to ‘double-check’ that nothing has been missed, but with time will hopefully lead to a faster, more convenient diagnostic service for the screening population.

Scoring Systems for Clinical Colon Capsule Endoscopy-All You Need to Know

In the constantly developing era of minimal diagnostic invasiveness, the role of colon capsule endoscopy in colonic examination is being increasingly recognised, especially in the context of curtailed endoscopy services due to the COVID-19 pandemic. It is a safe diagnostic tool with low adverse event rates. As with other endoscopic modalities, various colon capsule endoscopy scores allow the standardisation of reporting and reproducibility. As bowel cleanliness affects CCE’s diagnostic yield, a few operator-dependent scores (Leighton–Rex and CC-CLEAR scores) and a computer-dependent score (CAC score) have been developed to grade bowel cleanliness objectively. CCE can be used to monitor IBD mucosal disease activity through the UCEIS and the panenteric CECDAIic score for UC and CD, respectively. CCE may also have a role in CRC screening, given similar sensitivity and specificity rates to conventional colonoscopy to detect colonic polyps ≥ 10 mm and CRC. Given CCE’s diagnostic yield and reproducible clinical scores with high inter-observer agreements, CCE is fast becoming a suitable alternative to conventional colonoscopy in specific patient populations.