Segmentation of Lung Region in Computed Tomography (CT) Images

Segmentation of lung region in lung CT scan images is an important pre-processing
technique prior automatic detection and classification of emphysema disease. Well
segmented lung allows correct selection of region of interest (ROI) and thereby improve
the abnormality detection and classification of the lung. In this work, a lung
segmentation algorithm for CT images is proposed and evaluated. The proposed method
is uses Gaussian smoothing filter followed by thresholding to create binary mask for the
lung region. Formally, the binary mask will only select the lung region and zero the all
the regions surrounding the lung area. The binary mask can be set to either separate the
left and right lung or to show both lungs simultaneously. The algorithm is tested on a
database of lung CT scan images of emphysema patients. The database contains top,
middle and bottom sections of the lung. Evaluation of the algorithm using 39 middle
section lung CT scan images give 15.38% correct segmentation of the left & right lung.
With 39 top and 37 bottom section lung images, the algorithm give yy43% and 82.05%
correct segmentation of lung. These results show the good potential of the proposed
algorithm for segmentation of the lung region in CT images.