Machine Learning for Interpretation of Brain Images: A Detailed Analysis through survey

Abstract

Biomedical imaging systems aim to make detection, diagnosis and ease the treatment of ailments with minimum patient discomfort for the patient and maximum accuracy. In this century, the development of biomedical imaging goes far beyond the identification of morphological features such as the location of malignant cells and extends towards providing a helping hand in validating the clinical outcome and subsequent processes through Artificial Intelligence (AI). Human Brain is an important organ and continuously monitors all the systems of the human body. Cognitive neuroscience merges the experimental approaches of cognitive psychology with various methods to interpret mental activities supported by brain function. Brain imaging includes Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET) and functional Magnetic Resonance Imaging (fMRI). A brain tumor can be traced to a remote location, which may be inaccessible, and the obscurity lies in locating the size and shape of the tumor. Lack of skills in reading the scan, time-consuming interpretation and limits of human vision capability call for more advanced methodologies such as Artificial Intelligence. It is deduced that AI techniques can produce unprecedented accuracy from the fact that current machines are trained to visualize human organs with the accuracy of human experts and can produce a 3D map of the organ with exact boundaries. This paper presents a survey of techniques used for brain imaging with Artificial Intelligence taken into service.