Proposal of high-risk adenoma detection rate as an impactful, complementary quality indicator of colonoscopy

ABSTRACT BACKGROUND: The effectiveness of colonoscopy for colorectal cancer (CRC) screening depends on quality indicators, which adenoma detection rate (ADR) being the most important. Proximal serrated polyp detection rate (pSPDR) has been studied as a potential quality indicator for colonoscopy. OBJECTIVE: The aim is to analyze and compare the difference in ADR and pSPDR between patients undergoing screening colonoscopy and an unselected population with other indications for colonoscopy, including surveillance and diagnosis. METHODS: This is a historical cohort of patients who underwent colonoscopy in the digestive endoscopy service of a tertiary hospital. Out of 1554 colonoscopies performed, 573 patients were excluded. The remaining 981 patients were divided into two groups: patients undergoing screening colonoscopy (n=428; 43.6%); patients with other indications including surveillance and diagnosis (n=553; 56.4%). RESULTS: Adenoma detection rate of the group with other indications (50.6%) was higher than that of the screening group (44.6%; P=0.03). In regarding pSPDR, there was no difference between pSPDR in both groups (screening 13.6%; other indications 13.7%; P=0.931). There was no significant difference in the mean age (P=0.259) or in the proportion of men and women (P=0.211) between both groups. CONCLUSION: Proximal serrated polyp detection rate showed an insignificant difference between groups with different indications and could be used as a complementary indicator to adenoma detection rate. This could benefit colonoscopists with low colonoscopy volume or low volume of screening colonoscopies.

Download Full-text

Artificial Intelligence in Colonoscopy: Improving Medical Diagnostic of Colorectal Cancer

Frontiers in Health Informatics ◽

10.30699/fhi.v9i1.209 ◽

2020 ◽

Vol 9 (1) ◽

pp. 27

Author(s):

Stefanus Bernard ◽

Arli Aditya Parikesit

Keyword(s):

Artificial Intelligence ◽

Colorectal Cancer ◽

Survival Rate ◽

High Risk ◽

Detection Rate ◽

Adenoma Detection Rate ◽

Polyp Detection ◽

Adenoma Detection ◽

Study Results ◽

Medical Diagnostic

Introduction- Colorectal cancer (CRC) is a development of abnormal cells either in colon or rectum. CRC considered being the 3rd leading cause of death in 2018 only behind lung and breast cancer. It first arises during pre-cancerous stages called as polyps. The detection and removal of polyp is important to increase the survival rate of patient. Various method of polyp detection are available. However, only colonoscopy remains the gold standard in detection and removal of polyps. Several studies showed how Artificial Intelligence (AI) used in colonoscopy area particularly in detecting polyps, assessing physicians and predicting patient with high risk of CRC. The aim of this study is to describe the involvement of AI in colonoscopy and its impact in reducing the Materials and methods– Search for journal articles conducted between May and June 2016 from various resources including PubMed and Google Scholar. 6 research journals were reviewed and all the advantages and limitations were discussed throughout this study. Results– Various study showed that AI able to improve medical diagnostic of CRC in several ways, including in the improvement of adenoma detection rate (ADR) in terms of medical diagnostic, finding physicians associated with high Adenoma Detection Rate (ADR) and predicting patients with high risk of CRC. In addition, the use of AI in colonoscopy also associated with limitations including require large amount of datasets and advance computational resources in order to generate accurate output. Conclusion– The utilization of AI in colonoscopy shows how it able to improve the diagnosis accuracy and survival rate of patients associated with CRC despite several limitations that were identified during the study. However in the future, instead of allowing it to fully automatically conducting diagnosis, it still needs to be accompanied by physicians conducting the operation as there is no hundred percent perfect algorithms.

Download Full-text