Evaluating Performance of EEG Data-Driven Machine Learning for Traumatic Brain Injury Classification

Abstract : Objectives: Big data analytics can potentially benefit the assessment and management of complex neurological conditions by extracting information that is difficult to identify manually. In this study, we evaluated the performance of commonly used supervised machine learning algorithms in the classification of patients with traumatic brain injury (TBI) history from those with stroke history and/or normal EEG. Methods: Support vector machine (SVM) and K-nearest neighbors (KNN) models were generated with a diverse feature set from Temple EEG Corpus for both two-class classification of patients with TBI history from normal subjects and three-class classification of TBI, stroke and normal subjects. Results: For two-class classification, an accuracy of 0.94 was achieved in 10-fold cross validation (CV), and 0.76 in independent validation (IV). For three-class classification, 0.85 and 0.71 accuracy were reached in CV and IV respectively. Overall, linear discriminant analysis (LDA) feature selection and SVM models consistently performed well in both CV and IV and for both two-class and three-class classification. Compared to normal control, both TBI and stroke patients showed an overall reduction in coherence and relative PSD in delta frequency, and an increase in higher frequency (alpha, mu, beta and gamma) power. But stroke patients showed a greater degree of change and had additional global decrease in theta power. Conclusions: Our study suggests that EEG data-driven machine learning can be a useful tool for TBI classification. Significance: Our study provides preliminary evidence that EEG ML algorithm can potentially provide specificity to separate different neurological conditions.
 • Machine learning approaches such as support vector machine (SVM), decision tree (DT), artificial neural network (ANN) have also been used to classify TBI. • Most existing machine learning models use inputs from various brain imaging techniques to classify TBI patients. • In contrast, here, we directly use EEG for classification as the ultimate goal is to build a portable system which will be capable of detecting mTBI via EEG assessment in a field-based scenario.
 ? It also needs to be noted that almost all models published for TBI classification utilized supervised learning. ? However, one of the challenges in the detection and monitoring of TBI is the lack of an early and sensitive outcome measure. ? This restrains the performance of classifiers within our current knowledge breadth. McCrea et al. reported that an EEG-based algorithm could potentially be more sensitive than conventional neurocognitive assessment in monitoring the recovery from TBI. ? Our own study in mice also suggests EEG changes can be observed without an apparent neuroinflammatory reaction . ? Therefore, in the future, an unsupervised approach can be explored to mitigate this limitation
 ? An identical Conv-Pool combination repeats after that, giving the output to a batch-normalization layer then to 40 neurons of dense layer with L1 regularization equal to 0.001. ? Here, we use categorical cross-entropy with the Adam optimizer and early stopping of 50 epochs for training. ? Although other works proposed (for different tasks) deeper networks, avoiding manual feature extraction layers, yet our initial experiments showed leveraging decibel normalization method on the features significantly helps generalize the model . ? A large domain mismatch between train and test sessions makes this normalization method crucial.
 ? EEG has advantages of being noninvasive, easy-to-use, portable and cost effective. However, when applied to TBI research, EEG yields mixed results in the literature. ? Views on the clinical significance of EEG in TBI assessment are historically controversial . ? Studies have shown significant differences in EEG-based power spectra data between mild TBI and normal groups while other studies report no such distinction. ? Researchers have also evaluated post-TBI changes in connectivity and entropy .
Download DOC Download PPT

We have more than 145000 Documents , PPT and Research Papers

Have a question ?

Mail us : info@nibode.com