Image thresholding approaches for medical image segmentation - short literature review

Abstract

Medical image analysis is an invaluable tool in medicine. Different imaging modalities provide an effective means for mapping images that can feed machine and deep learning models which can significantly contribute to increase knowledge of diseased anatomy for medical research, being an important component in diagnosis and treatment planning. Accurate segmentation of medical images is a key step in the use of Artificial Intelligence as a tool to support physicians. Although there are many approaches to medical image segmentation, thresholding is one of the simplest methods that can be used, being of great interest to the scientific community. Over the years, several approaches to segmentation of medical images based on threshold methods have been proposed, which have evolved both in computational complexity, with shorter processing times, and in efficiency and accuracy of results. This manuscript summarizes the most common threshold-based approaches for medical image segmentation and discuss some of the methods and algorithms proposed in recent years, analyzing their advantages, disadvantages, and limitations.