Research And Application Of Medical Imaging Classification Technology Based On Bayesian Model

With the development of digital medical imaging technology, the medical imaging equipment is widely used in clinical medicine, and the computer-aided diagnosis application based on medical imaging technology also spread rapidly. Various medical imaging modes have also been an important objective basis of tracking disease, clinical medical diagnosis and teaching research and so on, and it also promotes the development of modern clinical medical diagnosis technology. Applying the classification technology to process and analyze the medical image, mine deep essential characteristic information. Thus it will assist medical clinical diagnosis, reduce the doctor’s workload and error rate for depending on clinical experience to diagnose. Besides, it has a higher value for the academic research, and the application scenarios are also extensive. The characteristics of medical image itself, such as various samples, high dimension and uneven sampling period, etc, which make medical image diagnosis recognition become a complex and nonlinear problem. Researchers have proposed the methods which are based on neural network, and the probability statistics, such as Bayesian Classification method etc.Based on the Bayesian classification, the researchers have put forward improvements in many aspects, including the most basic Naive Bayesian classification model NB and the AODE classification model etc. For the high real-time requirement of application scenario, researchers also have proposed the corresponding solution, AAPE classification model. When choosing the super parent, AAPE classification model itself does not consider selection order, and it may affect the classification results, thus the precision maybe lower during the early classification. Based on the disadvantange above, this thesis proposes AAPGE classification model which is based on the Genetic Algorithm and characteristic attributes weight. This model will weigh the calculation resources and the accuracy of classification results, and it enables these to achieve balance. Thus, with the request of real-time, the precision can also reach to a certain extent. The major contributions of the thesis are as follows:1. The thesis proposes the improved Bayesian Classification Model- AAPGE by introducing genetic algorithm and calculating characteristic attribute weight to AAPE classification model. Using mutual information and chi-square statistics information to calculate attribute weights to gain the calculated value which are separately considered as genetic algorithm’s fitness function, and using genetic algorithm to calculate attribute SPODE which is sorted. The result demonstrates that the effect of AAPGE’s classification model is better than the AAPE’s, and the accuracy is improved in early classification.2. The thesis designs and implements a medical imaging classification system prototype based on AAPGE classification model. The experimentresults indicate that the accuracy of AAPGE classification is more obvious than AAPE’s, and it ensures the classification’s real-time and accuracy to achieve a balance.