Brain tumor segmentation and classification using hybrid deep CNN with LuNetClassifier

Abstract

Brain tumour detection is essential for improving patient survival and prospects. This research work necessitates a physical examination with magnetic resonance imaging (MRI). As a result, computational algorithms are required for more accurate tumour diagnosis. Moreover, evaluating shape, boundaries, volume, size, segmentation, tumour detection, and classification remains difficult. To resolve these problems, hybrid deep convolutional neural network (DCNN) with enhanced LuNet classifier algorithm has been proposed for brain tumour detection. The main intention of the proposed approach is to locate the tumor and classify brain tumors as Glioma or Meningioma. For preprocessing, a Laplacian Gaussian filter (LOG) is used. A Fuzzy C Means with Gaussian mixture model (FCM-GMM) algorithm has been proposed for segmentation. To begin, use the extended LuNet algorithm to divide the data. A VGG16 extraction feature yields thirteen categorical features. Overall, the proposed method attempts to improve the performance of classifiers. The proposed LuNet classifiers are an excellent deep learning technique because it has low computational complexity, are inexpensive, and are simple to use even for those with little training experience. The simulated outcomes of the proposed algorithm compared to other conventional algorithms like SVM, Decision tree, Random forest, Alexnet, Resnet-50 and Googlenet classifier algorithm. The introduced hybrid approach achieves 99.7% accuracy. When compared to other existing algorithms, the proposed method outperforms them.