https://scholar.dgist.ac.kr/handle/20.500.11750/11754 2026-04-04T16:38:02Z 2026-04-04T16:38:02Z Seo, Youngho Hosseini, Zahra Kim, Kang Park, Jaebum Song, Tai Kyong Yu, Jaesok https://scholar.dgist.ac.kr/handle/20.500.11750/59961 2026-02-08T16:40:18Z 2025-09-16T15:00:00Z

Title: Spatio-Temporal Oriented Gradient (STOG) Filtering for Ultrasound Localization Microscopy: Preserving Slow and Fast Flow Components Author(s): Seo, Youngho; Hosseini, Zahra; Kim, Kang; Park, Jaebum; Song, Tai Kyong; Yu, Jaesok Abstract: Ultrasound Localization Microscopy (ULM) enables super-resolution vascular imaging but depends heavily on clutter filtering. Conventional Singular Value Decomposition (SVD) struggles to distinguish slow microvascular flows from static tissue, often suppressing diagnostically important signals. We propose a Spatio-Temporal Oriented Gradient (STOG) filter that exploits pixelwise gradient features in space and time. By combining co-occurrence and temporal elevation analysis, STOG separates both fast and slow flow components. Experiments with flow phantom and in vivo rabbit data demonstrated that STOG preserves microvascular structures missed by SVD, while maintaining overall vascular patterns. Despite minor tissue artifacts, STOG suggests a promising direction for gradient-based clutter filtering in angiogenesis imaging relevant to ischemic stroke prognosis. © 2025 IEEE.

2025-09-16T15:00:00Z Lee, Sangheon Guezzi, Nizar Jung, Dongkyu Seong, Hyojin Nam, Sangwoo Her, Taehoon Seo, Youngho Kim, Myeongchan Cho, Seonghyeon Choi, Suyoung Park, Jaebum Song, Tai-kyong Yu, Jaesok https://scholar.dgist.ac.kr/handle/20.500.11750/59392 2026-01-21T10:10:13Z 2025-09-15T15:00:00Z

Title: Learning-Based Design of Mismatched Filters via Unsupervised Deep Optimization for Coded Excitation Ultrasound Author(s): Lee, Sangheon; Guezzi, Nizar; Jung, Dongkyu; Seong, Hyojin; Nam, Sangwoo; Her, Taehoon; Seo, Youngho; Kim, Myeongchan; Cho, Seonghyeon; Choi, Suyoung; Park, Jaebum; Song, Tai-kyong; Yu, Jaesok Abstract: Coded excitation is widely used in ultrasound imaging to improve the signal-to-noise ratio (SNR), and, in particular, Barker-code-based pulse compression offers both high axial resolution and SNR gain. However, matched filtering yields elevated sidelobes that can degrade image quality, motivating the use of mismatched filters. Conventional mismatched filter designs rely on numerical optimization (e.g., minimizing integrated sidelobe level (ISL) or peak sidelobe level (PSL) under energy constraints) but are limited by fixed objectives, which restrict design flexibility. This work proposes a deep-learning-based, filter-to-filter framework that takes the transmit waveform as input and directly predicts mismatched filter coefficients via a multi-objective loss. The framework also integrates the transducer impulse response into the design process, enabling more precise mainlobe control and improved sidelobe suppression compared with a matched filter. Because the loss terms can be readily returned to different design goals, thus providing a flexible path for filter design in coded-excitation ultrasound systems.

2025-09-15T15:00:00Z Guezzi, Nizar Noman, Muhammad Lee, Sangheon Nam, Sangwoo Seo, Youngho Yu, Jaesok https://scholar.dgist.ac.kr/handle/20.500.11750/59391 2026-01-21T10:10:12Z 2025-09-17T15:00:00Z

Title: Deep Learning-Based Clutter Suppression for Single-Shot Ultrasound Flow Imaging Author(s): Guezzi, Nizar; Noman, Muhammad; Lee, Sangheon; Nam, Sangwoo; Seo, Youngho; Yu, Jaesok Abstract: Clutter filtering is a crucial step in ultrasound flow imaging for eliminating low-frequency signals arising from stationary or slowly moving tissue. Traditional clutter suppression techniques such as high-pass temporal filtering and singular value decomposition (SVD) rely on long temporal ensembles, making them unsuitable for real-time or single-frame processing. In this work, we introduce a deep learning-based method that enables clutter suppression from a single ultrasound frame—no angular compounding or ensembles required. We design an Attention U-Net architecture that incorporates spatial attention mechanisms to focus on flow-related features while attenuating clutter. Our model demonstrates strong clutter suppression and high structural similarity with ground truth filtered outputs. This work opens the door for real-time, single-frame blood flow imaging using deep learning.

2025-09-17T15:00:00Z Choi, Suyoung Nam, Sangwoo Guezzi, Nizar Lee, Sangheon Kim, Myeongchan Cho, Seonghyeon Seo, Youngho Park, Jaebum Song, Tai-kyong Yu, Jaesok https://scholar.dgist.ac.kr/handle/20.500.11750/59389 2026-01-21T10:10:14Z 2025-09-15T15:00:00Z

Title: 3D real-time ultrasound/photoacoustic anatomical/functional imaging platform for hemodynamic response assessment: A feasibility study Author(s): Choi, Suyoung; Nam, Sangwoo; Guezzi, Nizar; Lee, Sangheon; Kim, Myeongchan; Cho, Seonghyeon; Seo, Youngho; Park, Jaebum; Song, Tai-kyong; Yu, Jaesok Abstract: The hemodynamic response is a critical physiological indicator for assessing cardiopulmonary function and monitoring therapeutic efficacy in various diseases. Photoacoustic imaging (PAI), which combines the strengths of ultrasound and optical imaging, has emerged as a prominent noninvasive, non-ionizing modality for monitoring and evaluating hemodynamic responses. However, conventional Q-switched laser–based PAI systems, which are typically used for deep-tissue imaging at several centimeters beneath the skin, face inherent limitations due to the trade-off between pulse repetition frequency (PRF) and laser energy. These constraints hinder real-time performance and restrict imaging to primarily 2D cross-sectional views.In this study, we propose a 3D ultrasound/photoacoustic imaging platform that integrates an OPO-based DPSS laser system with a 2D matrix array transducer for hemodynamic response assessment. The platform employs customized multi-wavelength sequences to perform volumetric imaging and spectral unmixing, thereby enabling the evaluation of distinct optical components. To validate the system, we conducted phantom experiments simulating vascular structures. Through spectral unmixing analysis regarding two mixed inks (cyan & magenta), the platform displayed a cyan ink concentration ratio map based on the inks’ different optical absorption properties, demonstrating its functional imaging capability. Furthermore, real-time performance was confirmed by recording ink flow at 30 frames per second.The proposed system shows strong potential as a contrast-free vascular and hemodynamic imaging platform for both preclinical and clinical applications, and future studies will investigate its clinical feasibility in greater depth.

2025-09-15T15:00:00Z