Brain tissue segmentation using a convolutional neural network
Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
The segmentation of magnetic resonance (MR) brain images into white matter (WM), gray matter (GM) and cerebrospinal uid (CSF) play an importantrole in diagnostics and research of brain diseases. A multitude of approaches to automatic brain tissue segmentation have been reported in literature. In this paper, we evaluate the effectiveness of a deep convolutional neural network (CNN) by reducing the segmentation task to a classication problem, an area in which CNNs have proven effective. Our CNN implementation analyses the neighborhood around each voxel to correctly classify the tissue type. The network was trained and evaluated on the IBSR in accordance with the comparison methodology for segmentation algorithms presented in Valverde et al. The results show promising performance and potential for the voxel based CNN approach to brain tissue segmentation.
Place, publisher, year, edition, pages
IdentifiersURN: urn:nbn:se:kth:diva-168080OAI: oai:DiVA.org:kth-168080DiVA: diva2:814233