Semantic segmentation and quantification of lung lesions caused by SARS-CoV-2 infection by machine learning methods

Segmentation of chest and lung images plays a crucial role in supporting disease diagnosis, treatment planning, and surgical navigation. For accurate segmentation of lung images, this paper proposes a CT image segmentation algorithm based on classical methods and neural networks using machine learning methods. The neural network and machine learning approach has been thoroughly analyzed in comparison with standard segmentation methods (threshold detection, K-means clustering, histogram-based approach, and boundary detection). The experimental results show that the proposed algorithm is highly competitive. © 2022 Author(s).