Subspace Pattern Recognition Method For Brain Stroke Detection

Now with the rising of living standard,the number of stroke incidence has increased.Stroke has become the one killer beyond the threat to human life after cancer.Research shows that the detection of stroke timely will greatly reduce stroke damage rate,if we are able to detect stroke in time before the arrival of ambulance and detecting instrument.It is particularly important for the early diagnosis of stroke,it has also attracted the attention of scholars at home and abroad.Microwave detection is used to detect the stroke by using microwave technique,has the advantages of low cost,high effectiveness and good safety performance.The relevant theory research of microwave to detect brain medium which cause different reflection coefficient and microwave detection.Firstly,this paper introduces the basic theory of microwave detecting stroke,including the structure of the complex brain organization and the basic principle of microwave detection.In fact,imaging technology is the detection technology which microwave nondestructive testing technology combine with microwave and electromagnetic inverse scattering imaging algorithm.Microwave test system mainly consists of microwave circuit and data analysis environment to collect data analysis and processing under the MATLAB environment.The current microwave detection method based on microwave imaging,which scan the microwave signal by ultra-wideband antenna and generate excitation signals by modulated Gaussian signal.Aiming at the challenges of the small and complex structures of the human brain,we propose a stroke detection based on pattern recognition system,and design appropriate classifiers to separate the different types of training samples effectively and fast.Secondly,this paper mainly focuses on stroke detection and locates algorithm based on pattern recognition,pattern recognition is divided into supervised classification and unsupervised classification.In this paper,this paper focuses on the supervised recognition,feature extraction,selection and the design of the classifier is the core of the whole system,the purpose of pattern recognition is to find the interface between two types(or multiple classes)in the feature space.In general,pattern recognition can be divided into data acquisition and preprocessing,feature extraction and selection,finally to verify this classifier.Firstly,the samples are extracted by singular value decomposition(SVD),and then to calculate the inner product space distance dimension reduction method.The method can be used to distinguish the hemorrhagic patients from the healthy ones.Then we build the experimental platform,which mainly include the ultra-wideband antenna.Using the materials that have the same dielectric properties with the specific brain tissues that can construct the brain model,and then set the related parameters,such as the setting of the frequency,bandwidth,sampling,power etc...The microwave measurement module use the Rod Schwarz's(ZVL)two port vector network analyzer,the frequency range from 10 MHZ to 6GHZ,it has very high measurement accuracy,and can measure dual port S parameters,this paper mainly uses the S21 parameter.All of these components applied in our experiments constitute our stroke detection system.Finally,summarize the full text,and make a prospect.The microwave detection system is suitable for the diagnosis and treatment of the prohospital,which can improve the survival rate of patients with stroke,so it has great potential for early diagnosis and treatment.