Fire-fly based MKFCM Segmentation and Hybrid Feature Extraction for Lung Cancer Detection

Abstract

The most serious and broad infections considered lung disease that sets up a principal general wellbeing risky and has a high demise level. In this worry, appropriate division of lung tumor from X-beam, CT output or, MRI is the moving stone to accomplishing totally electronic analysis framework for lung disease location. With the advancement of innovation and attainable quality of information, the regarded time of a radiologist can be secured by methods for PC apparatuses for tumor division. This paper, to improve the Lung cancer segmentation and classification a new model is introduce. To overawed the existing segmentation limitations in this proposed system for lung nodes detectionModified kernel-based Fuzzy c-means clustering (MKFCM) technique is used. The proposed method segmentation includes two modules, the fire-fly clustering module and the MKFCM clustering module. For feature Extraction feature of this paper a (Gray-Level Co-Occurrence Matrix), Local binary patterns (LBP) and Histogram of oriented gradients (HOG) based hybrid system is used. To select the best feature fire fly base Feature Selection (FS) technique is used. For proposed Lung cancer classification long short-term memory (LSTM) classifier is used. The proposed system is also named as FF-MKFCM-FF-FS-LSTM system. Finally the performances are evaluated. From that analysis the proposed module provide 96.55% of segmentation accuracy and the proposed classification provides 98.95% of classification accuracy.