A detection approach for solitary pulmonary nodules based on CT images

It has been indicated that detection of pulmonary nodules plays an important role in diagnosing lung cancer in early-stage. In this paper, we propose an algorithm for detecting solitary pulmonary nodules automatically. Firstly, the algorithm implements prepared processing on original CT images and adopts adaptive iteration threshold twice to complete pulmonary parenchyma segmentation. Secondly, the experiment combines histogram analysis with compactness feature to obtain candidate nodules, and then achieves feature extraction for ROIs. Finally, SVM classifier is constructed on the basis of the extracted features to recognize true nodules and label them on original images. Experimental results indicate that our algorithm can not only achieve high accuracy and specificity but also can reduce the misdiagnosis, which is able to supply reference information with the radiologist detecting pulmonary nodules.