Bat-inspired Metaheuristic Convolutional Neural Network Algorithms for CAD-based Lung Cancer Prediction

In this real-world, lung cancer (LC) is the foremost reason for mortality in both mankind in the present time, with an inspiring figure of around five million deaths every year. Computer tomography (CT) can deliver valuable information when diagnosing lung illnesses. The chief goal of this work is to identify cancer nodules in the lungs from a given input image of the lungs and to organize LC and its harshness. To locate cancer nodules in the lungs, Fuzzy c-means (FCM) based segmentation is used. In this paper, a BAT optimization-based learning rate modified Convolutional Neural Network algorithm is introduced to effectively predict lung cancer. Additionally, to improve the proposed classification performance, input image is decomposed with support of the Discrete Wavelet Transform (DWT). With is used to decompose the image into four sub-bands, in such case we considered the Low (LL) band image. And then segmented images are split into two groups of images, which are used for the training and testing process. the proposed scheme has validated with the help of the LIDC-IDRI publically available dataset. They are studied by applying a convolutional neural network, and instantly trained neural network for predicting LC. In the end, the system efficiency is checked by using MATLAB tool to obtain the results of this model. In this experimentation, we achieved the accuracy of 97.43 % with a minimum classification error of 2.57 % in lung cancer prediction. This method is used to diagnose lung cancer correctly, and also this method may also overcome the previous drawbacks in the lung cancer diagnosis method.