Abstract
Real-time functional MRI is a non-invasive technique that enables the real-time analysis of brain activation patterns. Coupled with machine learning algorithms such as support vector machines (SVM), it can be used to identify different brain states in real-time. In this work, we developed a real-time brain state decoder system. In our implementation, image volumes were immediately reconstructed after acquisition, then transferred over the network to an analysis server where the images were immediately processed and the brain state decoded using pre-trained SVMs. We also explored the feasibility of using the system as a brain machine interface (BMI). For this, we scanned two participants and asked them to control the movement of an arrow displayed on a projection screen by matching their ongoing brain activity to a pre-defined activation pattern. Only a correct match, determined by the SVM, would move the arrow. The system attained an overall processing time per image volume that is less than the scan repetition time set at 2s. Moreover, participants were able to successfully control the arrow's movement demonstrating the feasibility of the system as a BMI.
