Diagnosis of Distant Metastasis of Lung Cancer Using Anarchic Society Optimization with Support Vector Machine

Abstract

The most widely recognized threatening tumours is the lung cancer one amongst the most preeminent disease and mortality which prompts significant risk to individuals’ wellbeing and life. Propelled lung malignant growth is probably prompt to produce comparing side effects in patients with extraordinary torment and life-threatening. Computed Tomography (CT) is one of the efficient solutions to investigate the distant metastasis plausibility for which system with computer aided diagnosis is desirable. So to diagnose distant metastasis of cancer in lung, the past framework is accomplished by Support Vector Machine (SVM) based classification for manual segmentation and the precision achieved is 89.09% for the classification. To enumerate large data sets, manual segmentation is time consuming, tedious and labour-intensive and hence achieving feasibility is not that easy. The desired prediction accuracy is attained by the techniques namely efficient segmentation method and feature selection methods. The proposed system utilizes Improved Markov Random Field (MRF) with SVM based classification, thereby improving the system performance. In this research work, the input is nothing but the CT lung images and adaptive median filtering is utilized for preprocessing technique. Improved MRF approach is one of the solutions to segment the pre-processed image after which CT images are used to extricate the clinical features and radiomic features. Anarchic Society Optimization (ASO) algorithm provides assistance for choosing the optimal features which in turn progresses the classification accuracy. In view of the selected feature, SVM classifier is utilized for cancer classification. The test outcome proves that the projected framework achieves improved performance compared to the existing frameworks parameters like correctness, accuracy and review.