Journal of Engineering Research
Abstract
Brain tumor early prediction is a critical task in medical imaging, as early detection and classification of tumors can significantly improve patient outcomes and treatment planning. In this study, we propose multi-classification models based on deep learning techniques for early prediction of brain tumors using magnetic resonance imaging (MRI) scans. Specifically, we investigate the effectiveness of Convolutional Neural Networks (CNN) in the You Only Look Once (YOLO) approach for an accurate classification of brain tumors into multiple classes based on their morphological characteristics. The proposed model is designed to extract spatial features from MRI images, capturing local patterns and structures indicative of different tumor types. Moreover, the model is employed to analyze sequential MRI images over time, capturing temporal dynamics and changes in tumor characteristics. Experimental results on a large dataset of labelled MRI scans demonstrate the effectiveness of the proposed multi-classification model in accurately predicting brain tumor types at an early stage. Comparative analysis and evaluation metrics such as accuracy, validation, and losses' curves versus epochs, in addition to a confusion matrix, are presented to assess the performance of each model. Results have indicated that YOLO v8 yielded 96% accuracy in training samples and up to 100% accuracy in testing samples.
Recommended Citation
Elgohr, Abdelrahman T.; Elhadidy, Mohamed S.; Elazab, Mahmoud Dr; Hegazii, Raneem Ahmed; and El Sherbiny, Moataz M.
(2024)
"Multi-Classification Model for Brain Tumor Early Prediction Based on Deep Learning Techniques,"
Journal of Engineering Research: Vol. 8:
Iss.
3, Article 3.
Available at:
https://digitalcommons.aaru.edu.jo/erjeng/vol8/iss3/3