Artificial intelligence-assisted linked color imaging colonoscopy improves adenoma detection rate compared to linked color imaging colonoscopy alone

Aims Integration of artificial intelligence (AI) into white light imaging colonoscopy improves the adenoma detection rate (ADR). Linked color imaging (LCI) colonoscopy also improves ADR compared to white light imaging. The aim of this study is to compare the ADR of AI-assisted LCI colonoscopy with LCI colonoscopy alone.

Methods Patients undergoing total colonoscopy for the investigation of positive fecal immunochemical tests were included and retrospectively reviewed. AI-assisted colonoscopy (CADEYE, Fujifilm, Tokyo, Japan) was introduced in March 2020 in our institution. Before the introduction of AI, LCI colonoscopy had been performed.

Results A total of 189 patients were divided into AI-assisted LCI (n=70) and LCI (n=119) groups. ADR was significantly higher using AI-assisted LCI compared with LCI alone (66% vs 50%, p=0.041). The mean number of adenomas detected per patient was slightly higher using AI-assisted LCI compared with LCI (1.2+-1.4 vs 1.0+-1.2, p=0.261). There were no significant differences between the two groups regarding age, gender, body mass index, total colonoscopy time, cecal intubation time, non-polyp withdrawal time or Boston Bowel Preparation Scale. Polyps were slightly smaller in AI-assisted LCI compared with LCI alone (mean 5.9+-2.8 vs 6.4+-4.4 mm, p=0.373). The proportion of polyps located in the proximal colon was similar in the two groups. In multivariate analysis, using AI-assisted LCI and advanced age (>60-year-old) were significantly associated with higher ADR.

Conclusions AI-assisted LCI colonoscopy significantly improves the ADR compared to LCI alone colonoscopy. This is the first study reporting the effectiveness of AI integration into LCI.

Publikationsverlauf

Artikel online veröffentlicht:
14. April 2023