Uncertain Reasoning Ontology Based Modeling For Mid Cognitive Impairment Diagnosis

Mild Cognitive Impairment (MCI) is a transitional state between the cognitive changes of normal aging and Alzheimer’s disease (AD). AD is an irreversible disease of the central nervous recession. If the treatment can be given in the MCI stage, it could be effective to reduce the incidence of AD. However, the existing diagnosis methods including clinical observation and questionnaire test class are greatly influenced by the doctor and the patients’ subjective factors. It is necessary to seek a relatively objective and effective auxiliary diagnosis method to make up for the shortage.Ontology has been used in medical diagnosis decision system as a modeling method for normalized description and management domain knowledge. Nevertheless, ontology can’t achieve the representation and reasoning to the uncertainty knowledge for lacking of the processing capacity knowledge. For the above problems, this paper basing on the small world properties of Functional Magnetic Resonance Imaging (fMRI) realized the auxiliary diagnosis for patients with mild cognitive dysfunction. This paper constructed standardization description model of the fMRI concept by using ontology and extended the representation and reasoning ability of uncertain knowledge by using the reasoning ability of Bayesian networks. First of all, this paper constructed fMRI ontology model basing on the functional magnetic resonance imaging (fMRI) domain knowledge, builded bayesian ontology model for uncertainty reasoning by extending the Ontology Web Language (OWL) method. Finally merged the two above Ontology models into the fMRI bayesian Ontology model. The FBO ontology model can not only solve the problem of sharing the fMRI domain knowledge, but also the language expression of ontology description to uncertain knowledge. Basing on bayesian inference algorithm, FBO model can make the auxiliary diagnosis effectively inference and decision to such uncertainty problems.To test and verify the feasibility and effectiveness of FBO ontology model, this paper combined the demand with the function of the auxiliary diagnosis of MCI to implement a FBO ontology model as the core of MCI auxiliary diagnose prototype system. This system selected the resting fMRI data of forty subjects from the ADNI public database for performance testing, then calculated the three kinds of characteristics including the path length, overall efficiency and the hub node by mapping the test data as the AAL template (AAL). The results showed that this model can effectively help doctors to diagnose MCI with the prediction rate up to90%.