Alam, Md. AshrafulReza, Md TanzimFarha, RamisaNuha, Nigar SultanaSakib, Syed NazmusRafi, Sowat HossainKhan, Md Sabbir2022-09-052022-09-0520222022-01ID 18101406ID 18101143ID 18101160ID 18101140ID 18101274http://hdl.handle.net/10361/17166Cataloged from PDF version of thesis.Includes bibliographical references (pages 37-39).This thesis is submitted in partial fulfillment of the requirements for the degree of Bachelor of Science in Computer Science, 2022.2D computer vision and activities related to medical image analysis are remarkably guided with the help of Convolutional Neural networks (CNNs) in recent years. Since a chief portion in the available clinical imaging data is in 3D, we are inspired to further develop 3D CNNs for seeking the advantage of greater spatial context. Despite the fact that many FCNs are previously worked on and built by using various approaches, current 3D approaches still rely on patch processing due to the utilization of GPU memory, which limits the incorporation of bigger context information for improved performance. Using efficient 3D FCNs in MRI images without any data loss would result in more efficient disease detections. In this paper, we propose an approach to an efficient 3D U-net segmentation technique for MRI Images using a lossless preprocessing of an MRI image dataset. Our proposal has the advantage of an impressive reduction of the required GPU memory for 3D Medical Image processing activities and that too, with an enhanced performance which is evaluated by the IoU (Intersection over Union) evaluation metric. Comprehensive experiment results performed with MICCAI BraTS’20 exhibit the viability of the presented strategy.39 pagesenBrac University theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission.3D CNNFCNs.3D-UnetSegmentationVolumetric medical images3D medical image processingBrain 3D MRIImage processing -- Digital techniques.Neural networks (Computer science)Thesis