Efficient deep learning approach for brain tumor detection and segmentation based on advanced CNN and U-Net

Abstrak

In this paper, we propose an innovative deep learning methodology dedicated to tumor detection and segmentation in medical images using convolutional neural networks (CNNs) and the U-Net architecture. The study emphasizes the importance of improving the quality and relevance of these features by employing advanced preprocessing methods. The subsequent development involves training a CNN model to achieve accurate tumor classification within the medical images. Among the various deep learning techniques proposed for medical image analysis, U-net-based models have gained significant popularity for multimodal medical image segmentation. However, due to the diverse shapes, sizes, and appearances of brain tumors, simple block architectures commonly used in segmentation tasks may not adequately capture the complexity of tumor boundaries and internal structures. The experimental results provide compelling evidence of the proposed approach's efficacy in accurately detecting and segmenting brain tumors. The results highlight the successful performance of the approach and its ability to achieve accurate tumor identification and segmentation.