Cited time in webofscience Cited time in scopus

Full metadata record

DC Field Value Language
dc.contributor.advisor 박상현 -
dc.contributor.author MIGUEL ANDRES LUNA ACEVEDO -
dc.date.accessioned 2019-10-02T16:04:16Z -
dc.date.available 2019-10-02T16:04:16Z -
dc.date.issued 2019 -
dc.identifier.uri http://dgist.dcollection.net/common/orgView/200000171495 en_US
dc.identifier.uri http://hdl.handle.net/20.500.11750/10752 -
dc.description.abstract In recent years, Convolutional Neural Networks (CNN) have been applied to a wide range of computer vision tasks thanks to their ability to learn informative features from data without human supervision. CNN models have been successfully utilized to segment and classify natural images where there are large datasets, nonetheless, many structures have been proposed to improve their performances. By contrast, in the medical field even though datasets are smaller than their natural images counterpart, CNN models have also over-perform previous methods and are becoming the first choice to tackle a variety of probleMaster. However, it is unclear how CNN structures are able to make a better use of the limited training data to generalize and produce reliable predictions. Therefore, in this study a set of CNN structures was analyzed on healthy brain tissue segmentation as well as brain abnormalities segmentation to understand the effect of the architecture design and the loss function used to train the model. The main contributions of the present thesis are the proposed CNN architecture that is capable of segmenting the brain across different tasks, the proposed loss function that increases the accuracy and reduces the number of false negatives and the segmentation quality analysis across different metrics that allow us to understand the advantages and disadvantages of a CNN model. -
dc.description.statementofresponsibility prohibition -
dc.description.tableofcontents I. INTRODUCTION - 1

II. BASIC CONCEPTS AND BACKGROUND - 2
1 MRI IMAGES - 2
1.1 T1 WEIGHTED SEQUENCES - 3
1.2 T1 POST CONTRAST - 4
1.3 T1 WEIGHTED INVERSION RECOVERY - 4
1.4 T2 WEIGHTED SEQUENCES - 4
1.5 FLAIR - 5
2 BRAIN MRI SEGMENTATION - 5
2.1 COMPUTER ASSISTED SEGMENTATION - 5
2.2 BRAIN MRI SEGMENTATION PROCESS - 6
2.3 TRADITIONAL SEGMENTATION METHODS - 8
2.4 DEEP LEARNING SEGMENTATION METHODS - 9
2.5 EVALUATION - 11
2.6 LIMITATIONS - 12

III. METHOD - 15
1 PROPOSED MODEL - 15
2 VARIATIONS OF PROPOSED MODEL - 17
2.1 Model A - 17
2.2 Model B - 17
2.3 Model C - 17
2.4 Model D - 17
2.5 Model E - 18
2.6 Model F - 18

IV. EXPERIMENTAL RESULTS - 20
1 BRAIN TUMOR SEGMENTATION - 20
1.1 BraTS 2018 challenge description - 20
1.2 BraTS experiments - 21
2 WHOLE BRAIN SEGMENTATION - 24
2.1 MR brain segmentation challenge 2018 description - 25
2.2 Brain segmentation experiment - 26
3 WMH SEGMENTATION - 27
3.1 WMH segmentation challenge 2017 description - 28
3.2 WMH segmentation experiments - 29

V. DISCUSSION AND CONCLUSION - 32
References - 33
-
dc.format.extent 43 -
dc.language eng -
dc.publisher DGIST -
dc.source /home/dspace/dspace53/upload/200000171495.pdf -
dc.title Patch wise 3D CNN with transition layer and normalized loss function for MRI brain segmentation -
dc.type Thesis -
dc.identifier.doi 10.22677/thesis.200000171495 -
dc.description.degree MASTER -
dc.contributor.department Robotics Engineering -
dc.contributor.coadvisor Jaeil Kim -
dc.date.awarded 2019-02 -
dc.publisher.location Daegu -
dc.description.database dCollection -
dc.citation XT.RM 미14 201902 -
dc.date.accepted 2019-01-30 -
dc.contributor.alternativeDepartment 로봇공학전공 -
dc.embargo.liftdate 2021-02-01 -
dc.contributor.affiliatedAuthor Kim, Jaeil -
dc.contributor.affiliatedAuthor ACEVEDO, MIGUEL ANDRES LUNA -
dc.contributor.affiliatedAuthor Park, Sanghyun -
dc.contributor.alternativeName 김재일 -
dc.contributor.alternativeName MIGUEL ANDRES LUNA ACEVEDO -
dc.contributor.alternativeName Sang Hyun Park -
Files in This Item:

There are no files associated with this item.

Appears in Collections:
Department of Robotics and Mechatronics Engineering Theses Master

qrcode

  • twitter
  • facebook
  • mendeley

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE