Diagnosis and Classification of Brain Tumors Based on Deep Learning Techniques رسالة ماجستير

Abstract

Brain tumor diagnosis through imaging tests is a crucial aspect of medical treatment planning. The most common imaging test for diagnosing brain tumors is Magnetic Resonance Imaging (MRI). Despite the expertise of medical practitioners, accurately classifying the type of tumor can be challenging due to the complexity of medical images. This is why the study of using deep learning algorithms to classify brain tumors is crucial. In this study, a dataset of MRI images of patients diagnosed with brain tumors was used to train deep learning models such as Convolutional Neural Networks (CNN), Visual Geometry Group-16 Model (VGG16), and Visual Geometry Group-19 Model (VGG19). The results of the first phase showed an accuracy of 95.40% for the CNN model and 90-93% for the other deep-learning models. To improve the accuracy, the second phase applied transfer learning to the VGG16 model, resulting in a 96.27% accuracy improvement. In the third phase, data augmentation techniques were used to balance the data and prevent over fitting, resulting in an accuracy improvement of 98.45%. In conclusion, this study highlights the potential of deep learning algorithms in accurately classifying brain tumors and supporting medical practitioners in determining the best treatment plan for patients. The results demonstrate the significance of AI in the medical field and the potential for further advancements in the future.