Lung nodules identification rules extraction with neural fuzzy network

In this paper, a neural fuzzy model is proposed to extract the diagnosis rules for detecting the pulmonary nodules. First, series of image processing techniques including thresholding, morphology closing, and labeling to segment the lung area and obtain the ROIs are employed. Then, three main features, circularity, size of area, and mean brightness are extracted from ROIs and the nodules are identified with diagnosis rules that are obtained by the neural fuzzy model. Twenty-nine clinical cases including 583 CT slice images (512 /spl times/ 512 /spl times/ 8 bits) are tested in our study. The detection rate of the proposed method is 89.3%, and the false positive is approximately 0.3 per image. This result demonstrates that our method improves the detection rate and reduces false positive compared to other approaches. This also implies potential of this system in clinical practice.