부산대학교 도서관

Brain cancer is so deadly that diagnosis accuracy will be required before brain surgery. Segmentation technology is important for medical imaging. CNN model is capable of searching for overlaps of necrotic, edematous, growing, and healthy tissue, it might be hard to get relevant information from the images. We propose a 3D U-Net network that was improved and trained with the T1ce weighted MRI scan to find these four areas. U-Net can set up many encoder and decoder routes that can be used to get information from images that can be use in different ways. This study, utilizes the GPU in training and to reduce the computing time of the 3D PSLNT method which is able to patch multichannel images and all the data patches created by the training images are known as label maps. Each patch is set for a test image and similar patches are taken from the dataset. The matching labels for each of these patches are then combined to produce an initial segmentation map for the test cases. Tests on T1 ce-weighted MRI scans show that our proposed model for brain tumor segmentation best performance