Khan, Hussain Nasir  <br/><br/>
Multimodal 3D-MRI Brain Tumor Segmentation via Modified 3D U-Net with Transformer / 
Hussain Nasir Khan 
 - 61p.  Soft Copy  30cm 
<br/><br/> Accurate brain tumor segmentation in 3D magnetic resonance imaging (MRI) scans is<br/>pivotal for medical diagnosis and treatment planning. This master’s thesis introduces an<br/>advanced approach for multimodal 3D-MRI brain tumor segmentation. Our methodology<br/>combines a modified 3D U-Net architecture with Transformer-based self-attention, and dilated<br/>convolution layers. By leveraging multimodal information from T1-weighted, T2-weighted, T1<br/>Contrast Enhanced (T1CE) Imaging and FLAIR MRI sequences within the BraTS 2023 dataset,<br/>our method significantly enhances segmentation precision. The modified 3D U-Net with<br/>Transformer and dilated convolution layers enables effective capture of both local and global<br/>contextual information, facilitating the identification of complex structures and long-range<br/>dependencies within volumetric MRI data. Thorough experimentation and evaluation, including<br/>comparisons with 3D U-Net and its variants, highlight the superiority of our proposed model in<br/>terms of brain tumor delineation accuracy. These findings emphasize the potential of this hybrid<br/>architecture to advance medical image analysis, providing substantial benefits to patient care. 
<br/><br/><label>Subjects--Topical Terms: </label><br/>MS Robotics and Intelligent Machine Engineering                                          
<br/><br/><label>Dewey Class. No.: </label>629.8