딥러닝을 활용한 의료 초음파 미세혈관 영상화 (Medical Ultrasound Microvascular Imaging using Deep Learning)

서지정보

· 발행기관 : 대한의용생체공학회
· 수록지 정보 : 의공학회지 / 45권 / 6호 / 409 ~ 416페이지
· 저자명 : 김선호, 박춘수, 이시열, 이동언, 유재홍, 김민우

초록

Clinical ultrasound is a powerful diagnostic tool that enables the non-invasive detection of various diseases without the risks associated with radioactive exposure. The addition of Doppler imaging enhances its capabilities by allowing the evaluation of blood flow, which is crucial for diagnosing vascular conditions. However, the accuracy of vascular imaging is often compromised by strong clutter signals, which interfere with the detection of blood flow sig- nals. While conventional clutter filtering techniques, such as Singular Value Decomposition (SVD), can effectively separate these signals, they are computationally intensive and may not perform well in real-time applications. Fur- thermore, detecting signals from microvessels is particularly challenging due to their low intensity, necessitating more advanced filtering techniques. In this study, we propose a novel clutter filtering approach based on a deep learn- ing framework for improve vascular imaging. Furthermore, it does not rely on the use of contrast agents, making it safer and more accessible for clinical use. By overcoming the limitations of existing techniques, this framework has the potential to significantly advance the field of vascular ultrasound imaging.

영어초록

Clinical ultrasound is a powerful diagnostic tool that enables the non-invasive detection of various diseases without the risks associated with radioactive exposure. The addition of Doppler imaging enhances its capabilities by allowing the evaluation of blood flow, which is crucial for diagnosing vascular conditions. However, the accuracy of vascular imaging is often compromised by strong clutter signals, which interfere with the detection of blood flow sig- nals. While conventional clutter filtering techniques, such as Singular Value Decomposition (SVD), can effectively separate these signals, they are computationally intensive and may not perform well in real-time applications. Fur- thermore, detecting signals from microvessels is particularly challenging due to their low intensity, necessitating more advanced filtering techniques. In this study, we propose a novel clutter filtering approach based on a deep learn- ing framework for improve vascular imaging. Furthermore, it does not rely on the use of contrast agents, making it safer and more accessible for clinical use. By overcoming the limitations of existing techniques, this framework has the potential to significantly advance the field of vascular ultrasound imaging.

참고자료

· 없음