The impact of AI applications in prostate segmentation on improving clinical diagnosis, and treatment: A review of the literature

Abstract

Prostate cancer is regarded as the second most common cancer in the world. Review of the studies that had been done on this topic for the years 2018-2020 by searching in Scopus, Science Direct, PubMed, and Google Scholar databases. Keywords used in this searching were medical image processing, prostate ultrasound image segmentation, fuzzy segmentation, CNN segmentation, and deep learning segmentation. The overall obtained articles were 4731, after the limitations of the search strategy, there were only 8 articles involved in this study. Findings showed the necessity of prostate segmentation and its role in the diagnosis and treatment improvement; furthermore, there are various approaches to segment prostate gland, but not all of them are suitable to use, due to the accuracy and time limitation. In conclusion, according to the findings of 4 articles, which mean 50% of the included studies, the results stated that using the CNN algorithm and its different approaches is the highest accuracy method that can be used for prostate segmentation.