Fuzzy Bat Algorithm based Segmentation and Mean Weight Convolution Neural Network (MWCNN) Classification for Lung Images

One of the biggest causes of non-accidental death is cancer. Globally, lung cancer has been confirmed to be the leading cause of death from cancer in men and women. The risk of death can be minimized with an initial diagnosis, so that the doctors may give the necessary care within a prescribed period. It was a challenging effort to locate the region of the field in Enhanced Particle Swarm Optimization Kernel Support Vector Machine (EPSOKSVM). The Fuzzy Bat Algorithm (FBA) with Mean Weight Convolution Neural Network (MWCNN) algorithm is proposed to settle this issue, in favour of identifying a area of Region of Interest (RoI) in the lung images in order to increase the certainty of classification. Medical images are decomposed into several layers in the proposed study to derive detailed visual information from various levels of scale. By using FBA, the lung images are then subdivided to identify RoI . With FBA objective fitness function, the interval between cancer and non-cancer areas are determined. Through the optimization process the fitness value is determined, and then the RoI area is segmented. Then, the non-negative sparse coding features acquired in various scales are incorporated to create a multi-scale function as the ultimate depiction for a medical image. Using, Mean weight computation and Average pooling function for the LIDC-IDRI database, MWCNN algorithm is used to achieve better results of classification certainty. The research findings reveal that the proposed FBA with MWCNN algorithm produces better results in the lung classification in terms of precision, recall, accuracy and F - measure values.