Volume 30, Issue 3, 2021
Advanced Lung Cancer Detection using Juxta-Pleural Nodule Extraction and Optimization by PSO Algorithms
Abstract
In this proposed work the diagnosis of Lung Cancer, for speedy and accurate analysis in the recognition of Lung Nodules by the radiologists for the early detection and increases the survival rate of the people. In the detection and classification of the Lung nodules, the segmentation process is considered to be the mandate requirement. The process of separating out the lung nodules from the other tissues is the segmentation. The traditional conservative methods used for the segmentation process does not yield accurate results and mostly depends on the features generated by the radiologists. In this proposed work, a computerized algorithm is designed with specified features for identifying the lung nodules and analyzing the nature of the segmented output with the ground truth images. Also the particle swarm based optimization algorithms are used for the improved detection mechanism. The advantageous feature of our proposed work is the detection of the juxta pleural lung nodules present very close to the wall of the lungs. The work also aims at reducing the time constraint of radiologists, since manual detection is time consuming and leads to false detection. The key goal of the proposed work is to obtain the accurate results by using the advanced contrast enhancement and the region growing segmentation algorithms. The performance is analyzed by using accuracy, specificity and sensitivity values. The above implementation is done for CT lung image database. On equating with the three algorithms, CPSO, WPSO and PSO, the accuracy of the malignant tumor detection and extraction is comparatively very much improved in case of CPSO with the enhanced accuracy of 95.80%. Also the algorithm has shown an improved accuracy of 90% in all the 20 sample images taken. The CPSO Algorithm shows an enhanced accuracy of 90% in the 4 out of 10 datasets.
Keywords
Median filter, Adaptive histogram equalization, Particle Swarm Optimization (PSO), Inertia Weight PSO, Chaotic PSO