In the high-dimensional feature selec tion process,the best feature subset generation and classifier parameter optimiz ation were concerned.A two stage optimization algorithm for high-dimensional fe ature selection based on bayesian rough set (BRS),genetic algorithm (GA) and cuc koo search (CS) was proposed.Firstly,analyzed the shape,gray and texture feature s of 3000lung tumor CT images,and extracted 104dimensional feature components to jointly quantify ROI.Then the two phase optimization was c arried out.(1)the attribute importance was analyzed from the global relative gai n function,and the fitness function was constructed by combining the weight of t he attribute reduction length ,gene encoding weight function and the attribute i mportance.The best feature subset was generated through genetic operation of sel ection,crossover and mutation,and the feature dimension was reduced without redu cing the accuracy of classification.(2) CS was used to optimize the parameters o f support vector machine (SVM).Finally,the feasibility and effectiveness of the algorithm were verified by experiments.Experimental results show that the algori thm effectively improves the ability of identifying benign and malignant lung tu mors and reduces the time complexity of the algorithm.