An improved algorithm for the automatic isolation of lungs in CT studies

The complexity of detecting pulmonary nodules has led to the development of Computer Aided Systems (CAD) that automate and reduce the cost of this task. The first phase of such systems usually consists in preprocessing the Computer Tomography (CT) scans, with the aim of segmenting the lungs and eliminating the elements that might interfere with the process. This paper presents an automatic method for the segmentation of lungs into three-dimensional pulmonary high resolution CT images. The proposed method has three main steps, that combine both 3D and 2D techniques. Firstly the trachea and the main airways are removed from the volume; then the lung region is segmented by grey-level thresholding, separating the right and left lungs if a junction is visible in the image, and the lung contour is smoothed; finally, a ”region growing” is applied using two seeds from each identified lung, avoiding as such the incorporation of other elements that do not belong to the lungs.