Identification Of Liver B-ultrasound Images In Computer-aided Diagnosis Systems

This paper is devoted to the study of liver B ultrasound image recognition in computer aided diagnosis system(CAD).The goal is to establish a computer aided diagnosis platform which can help clinicians to identify multi class liver B ultrasound images.The computer aided diagnosis system consists of four modules: selection of region of interest(ROI),image preprocessing,feature extraction and selection and classifier recognition.First of all,the liver area is drawn under the guidance of an experienced doctor,the regions of interest are extracted within the designated area,and then the obtained ROI are preprocessed,including image denoising and image enhancement.The obtained ROI is extracted and chosen features,at last the features are inputted to the designed classifier.In this paper,we focus on the study of the texture feature extraction and selection algorithm and the design of the classifier.In this paper,for the acquisition of the texture features,three aspects are considered: gray level,space and frequency,because of the diversity of liver lesions and there is no certain rules to follow,related to the first order gray statistics,gray level co-occurrence matrix,gray difference histogram and wavelet packet transform.By using the extracted texture features,three cases of liver B ultrasound images(normal,fatty liver,liver cirrhosis)can be classified and identified.However,some of these features may contain redundant information,which has a negative impact on the classification results.So a feature selection module is designed after the feature extraction.This paper introduces genetic algorithm to choose a discernible high optimal feature subset from the original feature set,in order to eliminate redundant features.Finally,the optimal feature subset is used as the input.On the design of classifier,the Adaboost theory is introduced to build a better classifier based on the traditional neural network.The final classification results are analyzed,and finally computer-aided diagnosis of liver B ultrasound images is designed.