Logo
Please use this identifier to cite or link to this item: http://20.198.91.3:8080/jspui/handle/123456789/1675
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorKar, Avijit-
dc.contributor.advisorDas, Sarit Kumar-
dc.contributor.authorSaha, Somojit-
dc.date.accessioned2022-11-14T10:24:31Z-
dc.date.available2022-11-14T10:24:31Z-
dc.date.issued2015-
dc.date.submitted2017-
dc.identifier.otherTC2745-
dc.identifier.urihttp://20.198.91.3:8080/jspui/handle/123456789/1675-
dc.description.abstractAbstract Reconstruction of cerebral cortex or its parametric representation from MR image is on one hand, a challenging problem in computational neuroanatomy, on the other hand its an extremely important tool in diverse applications. Complete parametric description of the cortical boundaries using model-free, low level image processing techniques produce inefficient results because of inherent biological intricacies and attributes of image acquisition. While many acceptable and encouraging results have been obtained using deformable model-based approach, a fully automatic and standard algorithm with accurate outcome is still an open area of research. The focus of this dissertation is to investigate hybridization of the existing deformable models with extensive use of the knowledge of neuroanatomy to meet the target of full automation with minimum computation and highly accurate cortical reconstruction for such applications as functional mapping or morphometric analysis. Shortcomings of the proposed models for cortical reconstruction in the available literature include varied amount of human interaction, approximations and assumptions, heuristic tuning of several parameters and above all, the questionability of true representation of the cortical surface. In this research, one of my main contributions is optimization of image acquisition parameters for improved image quality, reducing computational burden in post-hoc processing. My other contribution is a novel Advanced Anatomy Guided Hybrid Deformable (AAGHD) model for cortical reconstruction. There are four major contributing factors in this model. First, there is hybridization of different existing deformable models as well as hybridization of low level and high level processing techniques. Second, the model is fully automatic, starting with initialization of the deformable contour progressing through various stages even into the deepest sulcul folds. Third, a novel external force field has been designed overcoming the problem of partial volume effect, especially at narrow, deep sulcul folds. Fourth, the reconstructed cortical boundary converges completely with CSF/Gray matter interface i.e. the true cortical boundary. Our model has been validated on real MR images of brain from various 1.5 T MR scanners.en_US
dc.format.extentxv, 121p.en_US
dc.language.isoenen_US
dc.publisherJadavpur Univesity, Kolkata, West Bengalen_US
dc.subjectMR image of Brainen_US
dc.subjectAdvanced Anatomy Guided Hybrid Deformable (AAGHD) modelen_US
dc.subjectCSF/Gray Interfaceen_US
dc.titleDeformable model and its application in anatomy guided cortical reconstruction of MR Image of brainen_US
dc.typeTexten_US
dc.departmentJadavpur Univesity. Department of Computer Science and Engineeringen_US
Appears in Collections:Ph.D. Theses

Files in This Item:
File Description SizeFormat 
PhD thesis (Computer Sc. & Engg.) Somojit Saha.pdf6.35 MBAdobe PDFView/Open


Items in IR@JU are protected by copyright, with all rights reserved, unless otherwise indicated.