Lungs segmentation by developing binary mask

Lungs Segmentation from chest CT slices is a precursor for CAD applications. Most of the lungs segmentation methods are scanner dependent. We propose a fully automated machine independent method for segmenting lungs from CT images. The algorithm comprised of three main steps. In the first step, gray level threshold value has been selected by maximizing within class similarity. In the second step, binary mask has been developed using selected gray level threshold value and improved by morphological operations. In the third step, lungs have been segmented utilizing binary mask and original CT slice images. The method has been tested on data set of 25 slices collected from two different sources. Results have been compared with manually delineated lungs on CT images by a radiologist. Mean overlapping fraction, precision, sensitivity/recall, specificity, accuracy and F-measure have been recorded as 0.9929, 0.9962, 0.9966, 0.9997, 0.9995 and 0.9964 respectively.