Open AccessCOMPARATIVE STUDY OF MODERN NEURAL NETWORK ARCHITECTURES FOR MEDICAL IMAGE SEGMENTATION PROBLEMS
Author(s) -
А. Нагметова,
А. Алдош
Publication year - 2021
Publication title -
ķazaķstan-britan tehnikalyķ universitetìnìņ habaršysy
Language(s) - English
Resource type - Journals
eISSN - 2959-8109
pISSN - 1998-6688
DOI - 10.55452/1998-6688-2021-18-3-83-88
Subject(s) - artificial intelligence , convolutional neural network , computer science , segmentation , image segmentation , artificial neural network , image (mathematics) , computer vision , task (project management) , transformation (genetics) , pattern recognition (psychology) , perception , class (philosophy) , machine learning , psychology , engineering , biochemistry , chemistry , systems engineering , neuroscience , gene
Computer Vision is the area of Machine Learning that is responsible for machine perception of visual information. Image segmentation is a subfield of Computer Vision that solves the task of dividing a digital image into segments by their class label. One of the main problems in the subfield is the scarcity of data and the restoration of spatial information for the classified image. This article is a brief survey of current Biomedical Image Segmentation approaches, specifically Convolutional Neural Networks architectures and the morphological transformation for data augmentation.