Improvements to the Brain Tumor Segmentation and Classification System Using Convolutional Neural Networks

Abstract

The analysis of medical imaging for the diagnosis and treatment of brain tumors must include the segmentation of the tumor. The complexity of the brain's structure and the heterogeneity of the tumors make it difficult to segment brain cancers from magnetic resonance imaging (MRI) pictures. Deep learning approaches have recently demonstrated encouraging outcomes in the segmentation of brain tumors from MRI images. Due to its capability to handle high-resolution pictures and segment the entire tumor region, the U-Net model is one of them and is frequently utilized. For the diagnosis and planning of brain tumor treatments, accurate segmentation of brain tumors using multi-contrast MRI images is essential. Deep learning models including U-Net, PSPNet, DeepLabV3+, and ResNet50 have demonstrated encouraging outcomes in the segmentation of brain tumors. With the use of the BraTS 2018 dataset, we compare these models in this research. We assess the models' performance using a variety of measures, including the Hausdorff Distance (HD), the Dice Similarity Coefficient (DSC), and the Absolute Volume Difference (AVD), and we look into how data augmentation and transfer learning approaches affect the models' effectiveness. The findings demonstrate that the 3D U-Net model, which had a DSC of 0.90, HD of 10.69mm, and AVD of 11.15%, had the best performance. Similar performance was attained by the PSPNet model, which had a DSC of 0.89, HD of 11.37mm, and AVD of 12.24%. Performance was lower for the DeepLabV3+ and ResNet50 models, with DSCs of 0.85 and 0.83, respectively. The 3D U- Net and PSPNet models in particular saw a considerable improvement in performance thanks to the data augmentation strategies. The performance of all models, especially the DeepLabV3+ and ResNet50 models, was greatly enhanced by the transfer learning technique as well. The 3D U-Net model with data augmentation and transfer learning is suggested for brain tumor segmentation using multi-contrast MRI images based on the findings and analyses. The study shows the potential of deep learning models for segmenting medical images and emphasizes the significance of using the best model and optimization methods for the particular application.