Extraction Of The Image Characteristics Of Brain Function Based On Gray System Theory

Brain functional Magnetic Resonance Imaging (fMRI) is a combination of the digital imaging processing and functional magnetic resonance imaging, it is a new noninvasive technique and has very high spatial resolution and temporal resolution. It is suitable for the research of neuroscience and the advanced brain function. fMRI is a powerful tool to study the process of cognizing activity in the brain, but usually the work that extracting activation regions is carried out by hand. The disease diagnosing is also depended on the manual work and the eyes observation. So it is not only a large expenditure of the time and energy but also likely to make mistakes because of doctor's subjective judgement. Therefore, the study of how to classify the fMRI imaging automatically by computer will make great sense.Grey System Theory (GST) is a new method for the study of data problem which has property of small amount and uncertainty. For non classic regularity data such as non-steady, non-Gauss and non-white noise, GST has the obvious advantages than other processing methods which need statistical and transcendent regularity.In this paper, the algorithm of brain fMRI feature extraction based-on GST is studied, the main contribution is as follows:1. The imaging principle, characteristic and data processing methods of fMRI is analysed thoroughly;2. The concept, property and the range of application of Grey Modeling is also discussed in detail, especially the mechanism and methods of GM(1,1) modeling;3. Then the parameters of GM(1,1) model are acquired by Grey Modeling, the GM(1,1) model is used for the feature extracting of human brain fMRI imaging of different functional state;4. A novel method of brain fMRI imaging feature extraction is put forward and studied;5. With the actual imaging data as the research object, the grey model is constructed and used for the feature extracting of people in different hand-movement state;6. At the same time, in order to validate the algorithm, the feature parameters are classified with the KNN classifier and the result of classifying is acceptable.The results indicate that it is feasible and available for brain fMRI feature extraction with the grey modeling method and this method shows a new way to the brain fMRI recognition of the future.