Small-Bowel Capsule Endoscopy Optimising Capsule Endoscopy in Clinical Practice

Abstract

This thesis evaluated three aspects of the clinical application of small bowel cap- sule endoscopy (SBCE). The first section looked at the use of the patency capsule in clinical practice and evaluated its limitations. Our retrospective study demonstrated that while patency as- sessment reduces the risk of capsule retention to almost zero, it has its drawbacks due to the likely excessive false positive rate of the test as it stands, with functional patency confirmed in only 55.3% of patients having the test. We performed a single-centre pro- spective study of an extended-time protocol, which increased the confirmation of func- tional patency by 17%, a significant improvement. Still, there is likely room to improve this further. The second section of this work evaluated completion rates in capsule endoscopy. Completion rates in the literature are heterogeneous, with rates ranging from 64% to 96% (median 80%) [1] Real-time monitoring and prokinetics have been shown to be of value only in patients with risk factors for delayed gastric emptying. Still, they are time-con- suming and require the IV administration of prokinetics. Our group has demonstrated the efficacy and drawbacks of the standard approach where the completion rates (90.2 % vs 96.3 %) and rates of significant findings (37.7 % vs 32.5 %) were similar in our population with risk factors for delayed gastric emptying who received prokinetics with real-time monitoring in comparison to a control group [2, 3]. Peppermint water has been shown to aid gastric emptying, and in a pilot study, its possible benefit has been demonstrated as an aid to completion in those with risk factors for delayed gastric transit. Here we performed a single centre prospective open-label non-inferiority trial comparing orally administered peppermint water solution versus real time monitoring protocol as an aid to completion of SBCE in patients with risk factors for delayed gastric transit. We demonstrated completion rates comparable to real-time monitoring and a control group without risk factors for delayed gastric emptying. Com- pletion rates overall were high and were similar in the 3 groups 94.7% Peppermint vs 90.7% Real time monitoring vs 95.3% in a control group. All 3 groups met the European Society of Gastrointestinal Endoscopy (ESGE) minimum completion KPI of 80%, with the peppermint and control groups meeting the higher target of >/=95% [1]. Peppermint wa- ter solution given at the time of capsule ingestion in patients with risk factors for delayed gastric emptying is equivalent in terms of completion rates to the current standard pro- tocol involving the administration of prokinetics and has advantages, it is safe, well tol- erated and reduces patient time in the department. In the final section of this thesis, we looked at the use of an artificial intelligence- enabled capsule reading platform to aid in the analysis and reporting of small bowel cap- sules. Capsule reading is a time-consuming and often tedious process with average reading times in the literature ranging between 30–120 minutes [2]. It demands a high level of concentration without distractions to avoid missing pathology [3]. Our study of 40 small bowel procedures has shown; Overall diagnosis correlated 100% between AI and standard reading. In a per-lesion analysis, AI-assisted reading captured 1268 (98.1%, 95% CI 97.15–98.7) of lesions, while standard reading captured 1114 (86.2%, 95% confidence interval 84.2–87.9), P < 0.001. But most interestingly, mean reading time went from 29.7 minutes with standard reading to 2.3 minutes with AI-assisted reading (P < 0.001), for an average time saving of 27.4 minutes per study. This would be a game changer if further data confirms the software's superiority to human readers for exclusion of normal and duplicate images.