학술논문
Image Classification of Brain Tumors through Hybrid Learning
Document Type
Conference
Source
2024 Fourth International Conference on Advances in Electrical, Computing, Communication and Sustainable Technologies (ICAECT) Advances in Electrical, Computing, Communication and Sustainable Technologies (ICAECT), 2024 Fourth International Conference on. :1-7 Jan, 2024
Subject
Language
Abstract
Brain tumor, the well-known deadly medical condition where treatment sometimes fails to show results. Various treatments and medical diagnosis shows ineffective as the chances of this condition appearing again are more than curing them in first attempt. The abnormal growth of cells inside or surrounding brain causes these brain tumors which may differ depending on the type of tumor in each individual. The study proposes a hybrid model to classify the brain images. First the images are pre-processed with the help of Image resizing, Gray scale conversion and noise removal with the help of combined Weiner filter and Median filter. Then the Gaussian Mixture based histogram equalization are used to enhance the images. A Dense Convolutional neural network is used for extracting the Features of the images. An optimized ResNet101 model with improved Aquila optimization is used for classifying the brain tumor images. Performance of the model is compared with other studies in literature. The proposed model has obtained an accuracy of 99.01%, along with the various performance metrics as sensitivity of 98.60%, specificity of 98.01%, precision of 98%, F1 score of 98.44%.