Lesion Recognition Method of Liver CT Images Based on Random Forest

Abstract

Random forest algorithm has been intensively researched and developed in the field of machine learning, thanks to its considerable performance on classification. In terms of the identification of liver CT images, random forest algorithm is deployed to train and discover the characteristics of several common liver lesions through the usage of features vectors, such as image gray, texture, etc. This paper proposes an improved random forest algorithm based on feature selections. Concluding from experiment, the revised algorithm obtains a promising accuracy of classification.