A Novel Deep Learning Framework for the Detection of Brain Tumor

• Published in: 2023 3rd International Conference on Digital Futures and Transformative Technologies (ICoDT2) pp. 1 - 5
• Main Authors: Abbas, Muhammad Sohail, Jamil, Sonain
• Format: Conference Proceeding
• Language: English
• Published: IEEE 03.10.2023
• Subjects: AlexNet; brain tumor detection; Deep learning; FCM; Image segmentation; K-Means; Machine learning algorithms; Magnetic resonance; Magnetic resonance imaging; ResNet-50; Sensitivity; Support vector machines; SVM
• DOI: 10.1109/ICoDT259378.2023.10325786

Brain Tumor segmentation has been in focus for the past couple of years. Several researchers have developed state-of-the-art techniques to solve the problem. In recent years, people started working with deep learning for this problem because machine learning did not significantly improve after a stage. Most researchers used Magnetic Resonance Image (MRI) for brain tumor segmentation. The proposed methodology improves this segmentation and classification compared to the state-of-the-art techniques. The methodology works in three phases: In the first phase, segmentation uses two machine learning techniques, K-Means Clustering and Fuzzy C-Means (FCM) Clustering. In the second phase, we worked on feature extraction from our images using deep learning models AlexNet and ResNet-50 with small alterations. In the third phase, we used a Support Vector Machine (SVM) to classify the tumor as benign and malignant. We used two separate algorithms from Machine Learning for segmentation and two techniques from Deep Learning for classification to compare which works better in both cases. K-Means and FCM showed significant improvements with results in segmentation, and for classification, we got the accuracy of 96.6% and 98.3% for ResNet-50 and AlexNet, respectively.