This paper describes an experiment involving visual object fMRI brain data and the NeuCube  architecture. fMRI spatio- and spectro- temporal data (SSTD), apart from EEG, audio and video data, comprises both space and time information, that requires a specific and specialized architecture to process, interpret and visualize the data for better understanding and interpretation of the information it may carries. At the same time, any patterns can be better recognized and thus new knowledge that may be embedded within the pattern can be extracted. From the experiment with the case study of Haxby fMRI data, NeuCube has accomplished better accuracy in recognizing the brain patterns compared with the standard machine learning techniques (i.e. SVM and MLP). In addition, the NeuCube method assists deep learning of the SSTD and deeper analysis of the spatio-temporal characteristics and patterns in the fMRI SSTD.
- Evolving spiking neural networks
- functional Magnetic Resonance Imaging
- pattern recognition
- spiking neural networks