Computer Aided Diagnosis (CAD) system has been developed for the early detection of breast cancer, one of the most deadly cancer for women. The benign of mammogram has different texture from malignant. There are fifty mammogram images used in this work which are divided for training and testing. Therefore, the selection of the right texture to determine the level of accuracy of CAD system is important. The first and second order statistics are the texture feature extraction methods which can be used on a mammogram. This work classifies texture descriptor into nine groups where the extraction of features is classified using backpropagation learning with two types of multi-layer perceptron (MLP). The best texture descriptor as selected when the value of regression 1 appears in both the MLP-1 and the MLP-2 with the number of epoches less than 1000. The results of testing show that the best selected texture descriptor is the second order (combination) using all direction (0, 45, 90 and 135) that have twenty four descriptors.