Automated Brain Lesion Detection and Segmentation Using MR Images

Computer vision and machine learning allows the image data to be seen by a computer or machine as a person would see it. This is a complex concept for a computer to comprehend since computers do not understand the three-dimensional perspective as a person views and understands it. Computer vision has variety of applications in industry, medicine, surveillance systems, video analysis, robotic, and etc. Image segmentation is one of the most challenging topics in computer vision and machine learning. As an application of image segmentation in biomedical research is to localize some specific cells and tissues, e.g., tumor or stroke in magnetic resonance images (MRI). Medical image segmentation helps physicians to find these lesions more accurately, and it can be great source of information in emergency cases that specialist is not available. Therefore, it is an important process in computerized medical imaging. Automated segmentation of brain lesions in MRI is a difficult procedure due to the variability and complexity of the location, size, shape, and texture of these lesions. This study presents four algorithms for brain lesion detection and segmentation using MR images.

Autorentext

Nooshin Nabizadeh received her PhD in Electrical and Computer Engineering from University of Miami (UM) in May 2015. Her main research fields are machine learning, computer vision, signal and image processing, and data mining. She is currently working in Center for Computational Science in University of Miami as a research analyst.