Application Of Multi-label Support Vector Machine In X-RAY Lung Disease Detection

With the aggravation of environmental pollution and the increase of severe and extreme weather,the probability of lung infection is greatly increased,which is important for accurate judgment of lung diseases.In reality,the difference between the images taken by X-RAY is very small,and some lung diseases are not easily detected through X-RAY.The inexperienced doctors often have insufficient experience in the interpretation of X-RAY,resulting in wrong and missed judgment of X-RAY.The application of machine learning method in X-RAY detection will be able to mine the potential information,which will help to improve the situation of wrong and missed judgment,and further help doctors to make accurate judgment on lung diseases.Support vector machine has been applied in many fields including portrait recognition,text classification,handwritten character recognition,bioinformatics,et al.Support vector machine is much better than other machine learning algorithms in small sample training set.Since the optimization objective of support vector machine is to minimize structural risk rather than empirical risk,the problems such as over-fitting are avoided.In X-RAY lung examination,patients often suffer from several lung diseases at the same time,so X-RAY lung disease recognition is a multi-label problem in the field of machine learning,and the use of multi-label support vector machines for X-RAY lung disease recognition will avoid this problem.Using multi-label support vector machine for X-RAY lung disease identification will avoided this problem.In this thesis,multi-label support vector machine is applied to the detection of X-RAY lung diseases.The specific content includes:(1)A new multi-label support vector machine is designed.Since the traditional multilabel support vector machine does not consider the problems of intersection and overlap of feature spaces,which leads to the loss of classification accuracy,a multi-label parallel support vector machine is designed to solve this problem.(2)MLPSVM is applied to X-RAY lung disease identification.X-RAY lung images have many difficulties such as image noise and unobvious feature information.Histogram equalization is used to enhance the image,and the trained ResNet34 residual convolution neural network is used to extract the features of X-RAY lung images.(3)X-RAY lung disease auxiliary diagnosis system is designed and implemented.Based on the above work,an X-RAY lung disease auxiliary diagnosis system is designed and implemented.The system adopts C/S architecture to avoid difficulties such as high calculation required by training model.The system has been tested that it has certain stability and accuracy.