Student : MA Obenembot
About the student
I am a Computer Science Honors student at the University of Johannesburg with strong interests in AI, deep learning, and information network. I am passionate about tackling challenging projects, from machine learning to research based topics. With a problem-solving mindset, balancing technical skills with innovation in both academic and my personal pursuits.
About the Project
This project focuses on developing and evaluating a Convolutional Neural Network (CNN) model for brain tumor classification using MRI scans. Brain tumors represent a critical health challenge, and timely diagnosis is essential for improving patient outcomes. Traditional manual analysis of MRI scans is both time-consuming and prone to human error, which motivates the use of deep learning methods to assist radiologists with accurate and efficient decision support. The proposed model leverages the power of CNNs to automatically learn spatial and hierarchical features from medical images, reducing the need for manual feature engineering. By training the network on MRI images across multiple tumor categories i.e Glioma, Meningioma, Pituitary tumors, and Notumor the model is able to distinguish between tumor types and non-tumorous tissue. The architecture integrates convolutional, pooling, and fully connected layers, enabling it to capture both low-level texture patterns and higher-level semantic information critical for tumor classification. In addition to model design, the project explores strategies to enhance CNN performance, including normalization techniques, dropout for regularization, and data augmentation (gaussian noise and rotations ±10o) to improve generalization. Performance is evaluated using metrics namely accuracy, precision, recall, and F1-score to provide a comprehensive assessment of classification quality.
