Identifying Mild Traumatic Brain Injury Patients From MR Images Using\n Bag of Visual Words

Abstract

Mild traumatic brain injury (mTBI) is a growing public health problem with an\nestimated incidence of one million people annually in US. Neurocognitive tests\nare used to both assess the patient condition and to monitor the patient\nprogress. This work aims to directly use MR images taken shortly after injury\nto detect whether a patient suffers from mTBI, by incorporating machine\nlearning and computer vision techniques to learn features suitable\ndiscriminating between mTBI and normal patients. We focus on 3 regions in\nbrain, and extract multiple patches from them, and use bag-of-visual-word\ntechnique to represent each subject as a histogram of representative patterns\nderived from patches from all training subjects. After extracting the features,\nwe use greedy forward feature selection, to choose a subset of features which\nachieves highest accuracy. We show through experimental studies that BoW\nfeatures perform better than the simple mean value features which were used\npreviously.\n

Ähnliche Arbeiten