Title

Enhancement of Perivascular Spaces using 3D Convolutional Neural Network

Alternative Title

3차원 합성곱 신경망을 이용한 혈관영상 화질 향상

DGIST Authors

Park, Sanghyun ; Jung, Euijin ; Kim, Jaeil

Advisor

박상현

Co-Advisor(s)

Jaeil Kim

Issued Date

2019

Awarded Date

2019-02

Citation

Euijin Jung. (2019). Enhancement of Perivascular Spaces using 3D Convolutional Neural Network. doi: 10.22677/thesis.200000171501

Type

Thesis

Table Of Contents

Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

I. INTRODUCTION 1

II. Related Works 3
1 Spatial Domain Approach . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Transform Domain Approaches . . . . . . . . . . . . . . . . . . . . . . . 3
3 Learning Based Approaches . . . . . . . . . . . . . . . . . . . . . . . . . 4
4 Medical Image Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . 4

III. METHODS 6
1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 SRCNN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3 VDSR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4 Dense Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5 Densely Connected Dense Network . . . . . . . . . . . . . . . . . . . . . 9

IV. RESULTS 10
1 Data set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2 Evaluation Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3 Quantitative Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4 Qualitative Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5 Discussion for comparison networks . . . . . . . . . . . . . . . . . . . . . 15
6 Discussion for network depth . . . . . . . . . . . . . . . . . . . . . . . . . 18

V. CONCLUSION 20

References 21

URI

http://dgist.dcollection.net/common/orgView/200000171501
http://hdl.handle.net/20.500.11750/10757

DOI

10.22677/thesis.200000171501

Degree

MASTER

Department

Robotics Engineering

Publisher

DGIST

Show Full Item Record