We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Detection Brain Tumor Disease Using a Combination of Xception and NASNetMobile.
- Authors
Dishar, Hiba Kahdum; Muhammed, Lamia AbedNoor
- Abstract
Among Artificial Neural Networks (ANNs), Deep Convolutional Neural Networks (CNNs) have emerged as the most effective architecture for solving complex image-driven pattern recognition tasks. With the availability of large datasets and improvements in hardware technology, research on CNNs has accelerated, and several interesting deep CNN architectures have been proposed. In this study, we propose a novel CNN model based on transfer learning, which combines two state-of-the-art architectures, Xception and NASNetMobile, for image classification. The model takes images of specific dimensions or called "best windowing of images" as input and uses Xception and NASNetMobile to classify them into two categories. The outputs of these architectures are then concatenated using a concatenate layer, and a dropout layer is added to prevent overfitting problems in CNN. We evaluated the suggested model for a challenging MRI brain tumor dataset consisting of 3000 images, with 1500 images classified as normal and 1500 images classified as abnormal. Our results show that the suggested model produced excellent accuracy, demonstrating its effectiveness in enhancing the performance of CNNs for image classification tasks.
- Subjects
ARTIFICIAL neural networks; PATTERN recognition systems; IMAGE recognition (Computer vision); BRAIN tumors; BRAIN diseases; DEEP learning
- Publication
International Journal of Advances in Soft Computing & Its Applications, 2023, Vol 15, Issue 2, p325
- ISSN
2710-1274
- Publication type
Article
- DOI
10.15849/IJASCA.230720.22