A Study on Twinkling Artifact for Detecting Breast Microcalcifications
- 주제어 (키워드) Medical Ultrasound , Micorcalcifications , Twinkling Artifact
- 발행기관 서강대학교 일반대학원
- 지도교수 유양모
- 발행년도 2025
- 학위수여년월 2025. 2
- 학위명 석사
- 학과 및 전공 일반대학원 전자공학과
- 실제 URI http://www.dcollection.net/handler/sogang/000000079624
- UCI I804:11029-000000079624
- 본문언어 영어
- 저작권 서강대학교 논문은 저작권 보호를 받습니다.
초록 (요약문)
Breast cancer is characterized by high incidence and mortality rates, as well as poor prognosis in advanced stages. Effective screening techniques are therefore critical for its management, with X-ray-based mammography currently serving as the standard method. However, mammography has limitations, including radiation exposure and patient discomfort, driving interest in ultrasound-based approaches for early detection. Recent research has focused on utilizing the Twinkling Artifact (TA) to detect microcalcifications (MCs) in breast tissue, aiming to enhance sensitivity in early breast cancer diagnosis. Existing studies, however, have limited consideration to in-vivo environments where diverse noise sources are present. Moreover, the detection of TA has traditionally relied on examiners, posing another limitation. To address these challenges, this study proposes a machine learning-based method for the automatic detection and classification of TA. Additionally, a filter bank technique is utilized to enhance TA signals adaptively. The proposed method achieved classification accuracy exceeding 97% and demonstrated signal amplification performance above 110dB. Notably, applicability of the classifier was validated through in-vivo experiments, highlighting its potential for practical implementation. These results suggest that the proposed techniques can serve as robust and automated solutions for TA detection in clinical settings.
more목차
I. Introduction 6
A. Breast cancer and microcalcifications 6
B. Twinkling artifacts for microcalcification detection 7
C. Research motivation 8
II. Ultrasound Doppler system 9
A. CW Doppler system 10
B. PW Doppler system 13
C. Color Doppler system 16
III. Autocorrelation based Doppler classifier 17
A. Preliminary study 18
B. Autocorrelation based Feature extraction 19
C. Dataset organization and Classifier training 22
IV. Multi band Doppler imaging 26
A. Hypothesis 26
B. MBDI signal block 27
C. Filter bank 28
D. Signal modeling and weight optimization 29
E. Visualization 31
V. Experimental result 32
A. Experimental setup 32
B. Experimental results 35
VI. Discussion and Conclusions 43
A. Discussion 43
B. Conclusions 46
Reference 48
