Detail View
Denoising of virtual monoenergetic images in photon counting CT using statistically optimal weighting coefficients
WEB OF SCIENCE
SCOPUS
- Title
- Denoising of virtual monoenergetic images in photon counting CT using statistically optimal weighting coefficients
- Alternative Title
- 통계적 최적 가중 계수를 이용한 photon counting CT 가상 단일 에너지 영상의 노이즈 제거
- DGIST Authors
- Jiwoo Min ; Okkyun Lee
- Advisor
- 이옥균
- Issued Date
- 2026
- Awarded Date
- 2026-08-01
- Type
- Thesis
- Description
- PCCT, Sinogram, Material Decomposition, Denoising, Virtual Monoenergetic Image
- Abstract
-
물질 분해(material decomposition) 영상으로부터 virtual monoenergetic images (VMIs)를 합성하는 것은 photon-counting computed tomography (PCCT)의 주요한 장점 중 하나이다. 하지만 물질 분해 과정에서 발생하는 노이즈 증폭은 합성된 VMI의 영상 품질을 저하시키는 주요 요인이다. 본 연구에서는 VMI를 생성하기 위해 sinogram에 곱해지는 energy-dependent weighting coefficients를 최적화하여 spatial resolution과 noise texture를 보존하면서 VMI의 노이즈를 줄이는 새로운 기법을 제안한다. 각 sinogram 포인트에 대해 통계적으로 선택된 인접 (neighbors) 픽셀 값을 활용하여 물리적 제약 조건 하에서 최적의 계수를 산출한다. 산출된 계수들은 basis sinogram에 기존의 mass attenuation coefficients 대신에 적용된다. 실험 결과, 본 제안 기법은 기존의 다른 denoising 방법들과 비교하여 spatial resolution 및 noise texture의 저하를 최소화하면서도 VMIs의 노이즈를 효과적으로 억제함을 확인했다.|Synthesizing virtual monoenergetic images (VMIs) from material decomposed images is one of the significant advantages of photon-counting computed tomography (PCCT). However, noise amplification in material decomposition degrades the image quality of VMIs. In this study, we propose a noise reduction method in VMIs by optimizing energy-dependent weighting coefficients while preserving spatial resolution and noise texture. Statistically selected neighbors for each sinogram point are used to calculate the optimal coefficients with a physics-driven constraint. The coefficients are then applied to the basis sinogram before any noise reduction methods are performed. We demonstrate that the method effectively reduces noise in VMIs while minimizing the degradation of spatial resolution and texture compared to other denoising methods.
더보기
- Table Of Contents
-
I. Introduction 1
II. Methods 3
2.1 The ML-based estimator 3
2.2 Statistical iterative reconstruction (SIR) 4
2.3 Variation of the likelihood-based bilateral filter (LBF*) 4
2.4 The proposed method 5
III. Settings 8
3.1 Fan-beam simulation study 8
3.2 Cone-beam experimental study 10
3.3 Evaluation of the methods 12
IV. Results 14
4.1 Results of fan-beam simulation study 14
4.2 Results of cone-beam experimental study 19
V. Discussion and Conclusion 26
VI. Appendix A. Bias correction in CBCT 28
VII. References 29
국문요약 33
- URI
-
https://scholar.dgist.ac.kr/handle/20.500.11750/60783
http://dgist.dcollection.net/common/orgView/200001006786
- Degree
- Master
- Publisher
- DGIST
File Downloads
- There are no files associated with this item.
공유
Total Views & Downloads
???jsp.display-item.statistics.view???: , ???jsp.display-item.statistics.download???:
