Detail View

  Department of Robotics and Mechatronics Engineering Theses Master

Timestep-Synchronized Frequency-Aware Pseudo-3D Diffusion for Structurally Consistent CT Volume Generation

Citations

WEB OF SCIENCE

Citations

SCOPUS

Metadata Downloads

XML
Excel
DC Field Value Language
dc.contributor.advisor 문인규 -
dc.contributor.author Yeonwoo Seo -
dc.date.accessioned 2026-09-01T19:31:36Z -
dc.date.available 2026-09-01T19:31:36Z -
dc.date.issued 2026 -
dc.identifier.uri https://scholar.dgist.ac.kr/handle/20.500.11750/60780 -
dc.identifier.uri http://dgist.dcollection.net/common/orgView/200001014299 -
dc.description Diffusion Models,Pseudo-3D Generation,Medical Image Synthesis,Volumetric Consistency -
dc.description.tableofcontents Ⅰ. Introduction 1
1.1 Motivation and Problem Definition 1
1.2 Limitations of Existing Methods 1
1.3 Proposed Method and Contributions 2
1.4 Paper Organization 3

Ⅱ. Related Work 3
2.1 Medical Image Generation with Diffusion Models 3
2.2 Diffusion-Based Medical Image Synthesis and Structural Limitations 3
2.3 Existing Approaches for Inter-Slice Consistency 4
2.4 Anatomical Priors and Frequency-Aware Structural Guidance 4

Ⅲ. Preliminary 5
3.1 Denoising Diffusion Probabilistic Models 5
3.2 Timestep Embedding and Conditional Modulation 6
3.3 Frequency Domain Analysis 6
3.4 Coarse-to-Fine Diffusion Characteristics 6

Ⅳ. Method 7
4.1 Overall Framework Overview 7
4.2 Anatomical Template Generation 7
4.3 Pseudo-3D Multi-Conditional Representation 7
4.4 Conditional Diffusion Formulation 8
4.5 Network Architecture 9
4.6 Frequency-Aware Loss 9
4.7 Inference Stage 12

Ⅴ. Experiments 13
5.1 Dataset 13
5.1.1 Virtual Skeleton Database （VSD） 13
5.2 Data Preprocessing 14
5.2.1 Volume Cropping and Slice Selection 14
5.3 Implementation Details 14
5.3.1 Experimental Environment 14
5.3.2 Training Configuration 14
5.4 Evaluation Metrics 14
5.4.1 Segmentation-Based Anatomical Evaluation 15
5.4.2 Image Quality Evaluation 15
5.5 Ablation Study 15
5.5.1 Ablation Study Design 15
5.5.2 Quantitative Ablation Results 16
5.5.3 Qualitative Evaluation 18

Ⅵ. Conclusions 20

References 22		 -
dc.format.extent 32 -
dc.language eng -
dc.publisher DGIST -
dc.title Timestep-Synchronized Frequency-Aware Pseudo-3D Diffusion for Structurally Consistent CT Volume Generation -
dc.title.alternative 시간 단계 동기화 주파수 인지형 Pseudo-3D 확산 모델을 이용한 구조적 일관성 기반 CT 볼륨 생성 -
dc.type Thesis -
dc.identifier.doi 10.22677/THESIS.200001014299 -
dc.description.degree Master -
dc.contributor.department Department of Robotics and Mechatronics Engineering -
dc.date.awarded 2026-08-01 -
dc.publisher.location Daegu -
dc.description.database dCollection -
dc.citation XT.RM 서64 202608 -
dc.date.accepted 2026-07-21 -
dc.contributor.alternativeDepartment 로봇및기계전자공학과 -
dc.subject.keyword Diffusion Models,Pseudo-3D Generation,Medical Image Synthesis,Volumetric Consistency -
dc.contributor.affiliatedAuthor Yeonwoo Seo -
dc.contributor.affiliatedAuthor Inkyu Moon -
dc.contributor.alternativeName 서연우 -
dc.contributor.alternativeName Inkyu Moon -
dc.rights.embargoReleaseDate 2030-08-31 -
Show Simple Item Record

File Downloads

  • There are no files associated with this item.

공유

qrcode
공유하기

Total Views & Downloads

???jsp.display-item.statistics.view???: , ???jsp.display-item.statistics.download???: