Research On Medical Image Classification Method Based On Convolutional Neural Network

Modern hospital produces a large number of medical images every day and these medical data provide abundant data source for medical image mining.Due to the medical image preserved in cloud is chaotic,we should sort out them and get the medical image data suitable for mining before the image data from cloud of medical image is applied to the actual mining work.With the development of the data mining,many method of data mining have been applied to medical image classification but most of them make feature engineering,which extract the statistical characteristics of medical image,and take advantage of effective statistical learning method for classification.Recently,the research of deep learning method has made great progress,and some effective deep learning methods are naturally applied to medical image analysis.Convolutional is one of typical application on image classification of deep learning methods.Using the convolutional neural network for image classification not only improves the accuracy of image classification but also eliminates feature engineering of the traditional statistical method and greatly improve the efficiency of image classification.Therefore,this paper focus on the use of convolutional neural network for medical image classification.Firstly,this paper reviews the research status in the field of image classification at home and abroad,then the paper introduced SIFT features of image which has a strong distinction and produced the model of the bag of words,which shows more effective than other statistical models.And it also introduces the basic principle and application of the bag of words and SIFT.Furthermore,we present the basic principle and application in the field of image classification of deep learning and convolutional neural network.Finally,according to the characteristic of the traditional statistical learning classification method and the convolutional neural network method,we explore the new classification method which combines the advantages of the two methods.At the final part where we present our experimental results,we firstly compared the results of the two experiments which we make the classification with bag of words and convolutional neural network,it turns out that the convolutional neural network can not only eliminate the feature engineering but also more effective than the bag of words on medical image classification.The following excerpt produces that we combined the advantages of traditional statistical method and deep learning and proved it is more effective than either of the two in the field of medical image classification.