CLASSIFICATION OF NORMAL AND ABNORMAL LUNGS WITH INTERSTITIAL DISEASES BY RULE-BASED METHOD AND ARTIFICIAL NEURAL NETWORKS

We devised an automated classification scheme by using the rule-based method plus artificial neural networks (ANN) for distinction between n ormal and abnormal lungs with interstitial disease in digital chest ra diographs. Four measures used in the classification scheme are determi ned from the texture and geometric-pattern feature analyses. The rms v ariation and the first moment of the power spectrum of lung patterns a re determined as measures for the texture analysis. In addition, the t otal area of nodular opacities and the total length of linear opacitie s are determined as measures for the geometric-pattern feature analysi s. In our classification scheme with these mea sures, we identify obvi ously normal and abnormal cases first by the rule-based method and the n ANN is applied for the remaining difficult cases, The rule-based plu s ANN method provided a sensitivity of 0.926 at the specificity of 0.9 00, which was considerably improved compared to performance of either the? rule-based method alone or ANNs alone. Copyright (C) 1997 by W.B. Saunders Company.