Open this publication in new window or tab >>2026 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]
Speed of sound (SoS) is the longitudinal propagation speed of ultrasound waves in a medium, and it plays a central role in ultrasound imaging in several ways: Beamforming, which is the conversion of received time-domain echo signals into spatially resolved images, requires an assumed SoS. An inaccurate beamforming SoS leads to aberration artifacts, degrading the quality of conventional brightness-mode (B-mode) images and impairing other ultrasound modalities that rely on accurately beamformed data. SoS is also an emerging quantitative ultrasound biomarker that can inform about tissue composition and pathological state.
This thesis focuses on pulse-echo SoS estimation using conventional handheld ultrasound transducers. In particular, the thesis investigates displacement-based SoS estimation, where apparent shifts between images acquired with different transmit sequences are used to infer the underlying SoS. While displacement-based approaches are promising, their performance depends on several factors, including the accuracy of utilized imaging model, the choice of regularization, the quality of measurements, and the robustness of the reconstruction method.
This thesis addresses several of these challenges, developing accurate and robust pulse-echo SoS estimation methods and evaluating their utility in clinically relevant applications. The key methodological contributions include: an analytical global (single-value) SoS estimator based on transmission geometry; its extension to windowed SoS estimation in a region of interest; formulating plane-wave SoS imaging as a convolution problem for effectively learning an improved imaging model from data; and a model-based deep learning framework that combines plug-and-play reconstruction with a measurement-refinement module that simultaneously denoises measurements, inpaints missing ones, and compensates for simplifications in assumed imaging models. The proposed methods are successfully evaluated in breast ultrasound applications through breast density classification with our global SoS estimation and through breast lesion characterization with local SoS imaging. Overall, this thesis contributes to the development of robust pulse-echo SoS imaging methods and supports the continued translation of SoS toward a practical quantitative biomarker in clinical ultrasound.
Place, publisher, year, edition, pages
Uppsala: Acta Universitatis Upsaliensis, 2026. p. 79
Series
Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology, ISSN 1651-6214 ; 2694
Keywords
Beamforming, Inverse problems, Image reconstruction, Model-based deep learning, Breast density, Breast cancer
National Category
Medical Imaging Radiology and Medical Imaging Signal Processing Cancer and Oncology Computer Vision and Learning Systems Computer graphics and computer vision
Research subject
Computerized Image Processing
Identifiers
urn:nbn:se:uu:diva-588526 (URN)978-91-513-2891-1 (ISBN)
Public defence
2026-09-04, Room 2001, Ångströmlaboratoriet, Regementsvägen 10, Uppsala, 09:15 (English)
Opponent
Supervisors
2026-08-172026-06-212026-08-17