The aim of this study was to develop a screening system of chest radiographs of miners with pneumoconiosis. Chest radiographs were of coal mine or silica dust exposed miners participating in a health screening program. A total of 236 regions of interest (ROI) (166, 49, and 21 with profusions of category (shape and size) 0, 1(q), and 1(r), respectively) were identified from 74 digitized chest radiographs by two B-readers. Two different texture feature sets were extracted: spatial gray level dependence matrices (SGLDM), and gray level differences statistics (GLDS). The nonparametric Wilcoxon rank sum test was carried out to compare the different profusion categories versus that of profusion 0 (normal). Results showed that significant differences exist (at a=0.05) between 0 versus 1(q), and 0 versus 1(r) for 14, and 12 texture features respectively. For the screening system, the self-organizing map (SOM), the backpropagation (BP), and the radial basis function (RBF) neural network classifiers, as well as the statistical k-nearest neighbour (KNN) classifier were used to classify two classes: profusion 0 and profusion 1(q and r). The highest percentage of correct classifications for the evaluation set (116 and 20 cases of profusion 0 and 1(q and r) respectively) was 75% for the BP classifier for the SGLDM feature set. These results compare favorably with inter- and intra-reader variability.