Abstract: The quantitative relation between physical features and computerized radiography (CR) parameters is analyzed with artificial neural network to develop a 0-1 integer planning model based on physical feature parameters, tube voltage, tube current· exposure time and CR scoring according to the satisfiability of relevant photos for body indices. Grouped filtering method was used to simplify the solution space, then the optimized model is solved by genetic algorithm. The result of pleurography (PA) indicates that the conclusion drawn by this CR optimization method, is compatible with experts' long-time clinical experience and that both the artificial neural network and genetic algorithm play the role in CR and solving actual problems. In addition, this method can be used to analyze the CR photos of the rest of body, and extended to other X-ray photographic equipment.