Image processing- based Lung TumorDetection and Classification using 3D MicroCalcification of CT Images

Abstract

Lung cancer is deadly of all tumor disease with a high death rate as the identification of lung
disease at the postponed stage is risky. Hence, the early prediction of lung disease and its treatment is vital
to increase the recovery rate. The pattern recognition model based on the micro-calcification of lung CT
images aids to classify the lung lesion disease using its texture and statistical features. The features selected
are coefficient of reflection and density of mass for the binned lung CT image physical feature
measurement that aids in identifying the malignant nodule. Then, the thresholding method is applied with
the three-dimensional (3D) sectional Region of Interest (ROI) using the material dimensions. Thus, the
lung lump dimension with its physical and statistical features are analyzed using 100 suspected images with
ten normal images. This model includes an SVM classifier for the classification of normal and cancer
images, exhibiting 98% of accuracy for the proposed system.