A5016658563- Ultrasound Imaging and Elastography
- Medical Image Segmentation Techniques
- Photoacoustic and Ultrasonic Imaging
- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Medical Imaging Techniques and Applications
- 3D Shape Modeling and Analysis
- Ultrasound and Hyperthermia Applications
- Domain Adaptation and Few-Shot Learning
- Ultrasonics and Acoustic Wave Propagation
- Generative Adversarial Networks and Image Synthesis
- Medical Imaging and Analysis
- Advanced MRI Techniques and Applications
- Cell Image Analysis Techniques
- Soft Robotics and Applications
- Advanced Neural Network Applications
- Advanced Radiotherapy Techniques
- Computer Graphics and Visualization Techniques
- Shoulder Injury and Treatment
- Advanced Image Processing Techniques
- Surgical Simulation and Training
- Flow Measurement and Analysis
- Multimodal Machine Learning Applications
- Elasticity and Material Modeling
- Advanced X-ray and CT Imaging
ETH Zurich
2016-2025
Uppsala University
2021-2025
Informa (Sweden)
2024-2025
École Polytechnique Fédérale de Lausanne
2014-2021
IBM Research - Zurich
2021
Weatherford College
2021
Human Computer Interaction (Switzerland)
2016
Board of the Swiss Federal Institutes of Technology
2011-2016
University of British Columbia
2005-2012
Spatial regularization is essential in image registration, which an ill-posed problem. Regularization can help to avoid both physically implausible displacement fields and local minima during optimization. Tikhonov (squared l2 -norm) unable correctly represent non-smooth fields, that can, for example, occur at sliding interfaces the thorax abdomen time-series respiration. In this paper, isotropic Total Variation (TV) used enable accurate registration near such interfaces. We further develop...
Cancer diagnosis, prognosis, and therapy response predictions from tissue specimens highly depend on the phenotype topological distribution of constituting histological entities. Thus, adequate representations for encoding entities is imperative computer aided cancer patient care. To this end, several approaches have leveraged cell-graphs, capturing cell-microenvironment, to depict tissue. These allow utilizing graph theory machine learning map representation functionality, quantify their...
Abstract Dysfunctional extracellular matrices (ECM) contribute to aging and disease. Repairing dysfunctional ECM could potentially prevent age-related pathologies. Interventions promoting longevity also impact gene expression. However, the role of composition changes in healthy remains unclear. Here we perform proteomics in-vivo monitoring systematically investigate (matreotype) during C. elegans revealing three distinct collagen dynamics. Longevity interventions slow stiffening prolong...
Variations in the shape and appearance of anatomical structures medical images are often relevant radiological signs disease. Automatic tools can help automate parts this manual process. A cloud-based evaluation framework is presented paper including results benchmarking current state-of-the-art imaging algorithms for structure segmentation landmark detection: VISCERAL Anatomy benchmarks. The implemented virtual machines cloud where participants only access training data be run privately by...
Abstract Sinusoidal endothelial cells and mesenchymal CXCL12-abundant reticular are principal bone marrow stromal components, which critically modulate haematopoiesis at various levels, including haematopoietic stem cell maintenance. These subsets thought to be scarce function via highly specific interactions in anatomically confined niches. Yet, knowledge on their abundance, global distribution spatial associations remains limited. Using three-dimensional quantitative microscopy we show...
Despite many uses of ultrasound, some pathologies such as breast cancer still cannot reliably be diagnosed in either conventional B-mode ultrasound imaging nor with more recent elastography methods. Speed-of-sound (SoS) is a quantitative biomarker, which sensitive to structural changes due pathology, and hence could facilitate diagnosis. Full-angle computed tomography (USCT) was proposed obtain spatially-resolved SoS images, however, its water-bath setup involves practical limitations. To...
Explainability of deep learning methods is imperative to facilitate their clinical adoption in digital pathology. However, popular and explainability techniques (explainers) based on pixel-wise processing disregard biological entities' notion, thus complicating comprehension by pathologists. In this work, we address adopting entity-based graph explainers enabling explanations accessible context, a major challenge becomes discern meaningful explainers, particularly standardized quantifiable...
Abstract Objectives The aim is to assess the feasibility and accuracy of a novel quantitative ultrasound (US) method based on global speed-of-sound (g-SoS) measurement using conventional US machines, for breast density assessment in comparison mammographic ACR (m-ACR) categories. Materials methods In prospective study, g-SoS was assessed upper-outer quadrant 100 women, with 92 them also having m-ACR by two radiologists across entire breast. For g-SoS, ultrasonic waves were transmitted from...
Ultrasound (US) beamforming is the process of reconstructing an image from acquired echo traces on several transducer elements. Typical approaches, such as delay-and-sum, perform simple projection operations, while techniques using statistical information also exist, e.g., adaptive, phase coherence, delay-multiply-and-sum, and sparse coding approaches. Inspired by feasibility success inverse problem (IP) formulations in reconstruction problems, computed tomography, we herein devise IP...
Purpose Compensation for respiratory motion is important during abdominal cancer treatments. In this work we report the results of 2015 MICCAI Challenge on Liver Ultrasound Tracking and extend 2D to relate them clinical relevance in form reducing treatment margins hence sparing healthy tissues, while maintaining full duty cycle. Methods We describe methodologies estimating temporally predicting liver from continuous ultrasound imaging, used ultrasound‐guided radiation therapy. Furthermore,...
Deformable Image Registration (DIR) of MR and CT images is one the most challenging registration task, due to inherent structural differences modalities missing dense ground truth. Recently cycle Generative Adversarial Networks (cycle-GANs) have been used learn intensity relationship between these 2 for unpaired brain data. Yet its usefulness DIR was not assessed. In this study we evaluate performance thoracic abdominal organs after synthesis by cycle-GAN. We show that geometric changes,...
For beamforming ultrasound (US) signals, typically a spatially constant speed-of-sound (SoS) is assumed to calculate delays. As SoS in tissue may vary relatively largely, this approximation cause wavefront aberrations, thus degrading effective imaging resolution. In the literature, corrections have been proposed based on unidirectional estimation or computationally-expensive posteriori phase rectification. paper we demonstrate direct delay correction approach for US beamforming, by...
Image-guided radiation therapy can benefit from accurate motion tracking by ultrasound imaging, in order to minimize treatment margins and radiate moving anatomical targets, e.g., due breathing. One way formulate this problem is the automatic localization of given tracked landmarks throughout a temporal sequence. For this, we herein propose fully-convolutional Siamese network that learns similarity between pairs image regions containing same landmark. Accordingly, it localize thus track...
Purpose: Optical coherence elastography (OCE) is a promising technique for high-resolution strain imaging in ocular tissues. A major strain-inducing factor the eye intraocular pressure (IOP), with diurnal physiological fluctuations reaching up to 5 mmHg. We study herein low-amplitude IOP modulation assess local corneal patterns. Methods: Ex vivo porcine eye-globes were adjusted an initial of 15 and subsequently 25 Corneal was induced by two subsequent cycles, which first increased then...