A model based segmentation approach for lung segmentation from chest computer tomography images

Segmentation of pathological lung regions from the body regions is the most challenging part in any Computer Aided Diagnosis (CAD) system due to the presence of Pathological Bearing Regions (PBR) which lies on the lung’s periphery. Here, a unique pathological lung segmentation method called reference-model based segmentation that uses shape property of human lung is proposed. This method trounces the difficulties in segmentation from traditional approaches by examining the shape knowledge of lung. The proposed segmentation approach constructs a reference lung model from input slices using a novel Sampling Lines Algorithm (SLA) and extracts the shape features. The segmentation work is validated using dataset consisting of Digital Imaging and Communications in Medicines (DICOM) standard chest CT images of seven patients from cancer institute Chennai, nine patients from Gemini Scans, Chennai and fourteen patients from Lung Image Data-base Consortium image collection (LIDC-IDRI). The segmentation method’s performance is analyzed against widely used segmentation methods namely Graph Cuts (GC), Region Growing (RG), Active Contour and Flood Fill in terms of accuracy, specificity, sensitivity, overlap score, Jaccard index, and also Dices similarity coefficients (DSC). The numerical outcomes specify that the proposed work attains an improved result against widely used segmentation techniques.