Abstract

Lung Lesion Parenchyma Segmentation Algorithm for CT Images

Author(s): R. Indumathi*, R. Haripriya and S. Akshaya

Segmentation is a precursor for most medical image analysis and classification in Computer Aided Diagnosis (CAD). The CAD of lung CT generally segments the area of interest (lesions), which enables the physician to diagnose the disease. This project presents a fully automatic algorithm for segmenting the lung lesion parenchyma. It is based on Threshold processing and Mathematical Morphology, where Threshold is done by Absolute Threshold Value Binarization and Boundary Detect Algorithm. Further Mathematical Morphology is done by Lung Parenchyma Repair Algorithm, based on horizontal scanning which repairs the lung parenchyma boundary more accurately. Finally, it uses area property of the connected components for separating the cancerous cells (lesions) from non-cancerous tissues.