Use of artificial intelligence in diagnostic cystoscopy of bladder cancer

Abstract

Background. At the current stage of science and technology development, artificial intelligence (AI) is being actively developed and gradually introduced into the healthcare system.Aim. To perform a literature review to assess the diagnostic value of AI in the detection of bladder cancer at the cystoscopy stage.Materials and methods. We carried out a bibliographic search of articles in Medline and Embase databases using the keywords “artificial intelligence”, “cystoscopy”, “TURBT”.Results. Automated image processing based on AI can improve the accuracy of cancer diagnosis during cystoscopy. According to the studies presented in the review, the sensitivity of AI system for the detection of bladder cancer via cystoscopy can reach 89.7–95.4 %, while its specificity is 87.8–98.6 %, which exceeds the diagnostic capabilities of standard cystoscopy in white light, the sensitivity and specificity of which, according to recent investigations, are approximately 60 and 70 %, respectively. Despite the promising results of these studies, modern science is currently at the stage of developing and evaluating the performance of various AI methods used to analyze cystoscopy images. To date, it would be premature to introduce and widely use these technologies in healthcare, since there are no prospective clinical studies to assess the effectiveness of AI systems in diagnostic cystoscopy and transurethral resection of bladder cancer.Conclusion. Few studies show that AI-based cystoscopy is a promising approach to improvement of the quality of medical care for bladder cancer. Further research is needed to improve the diagnostic capabilities of AI and introduce the obtained technological data into clinical practice.