Study of automatic detection algorithm for lung nodules

This paper studies a computer-aided detection (CAD) algorithm for automatically detecting lung nodules in CT images. Firstly, the lung areas are segmented from CT images. Then, the regions of interest (ROIs) including nodules and blood vessels in lung areas are extracted by thresholding techniques based on the distribution of the gray levels for the images. After extracting the features to distinguish between nodules and blood vessels, a rule-based classification method is used to classify the ROIs. Finally, the lung nodules can be detected by utilizing the features. The algorithm had been tested with 232 CT images. Experimental results demonstrated that a high detection rate of 85% could be achieved when the algorithm was applied.