A5064865521- Photoacoustic and Ultrasonic Imaging
- Ultrasound Imaging and Elastography
- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- Surgical Simulation and Training
- Medical Imaging and Analysis
- Aortic aneurysm repair treatments
- Digital Imaging for Blood Diseases
- Flow Measurement and Analysis
- Soft Robotics and Applications
- Cardiac, Anesthesia and Surgical Outcomes
- Robotics and Sensor-Based Localization
- Ultrasonics and Acoustic Wave Propagation
- Advanced MRI Techniques and Applications
- Advanced X-ray and CT Imaging
- Medical Imaging Techniques and Applications
- Colorectal Cancer Screening and Detection
- Robotic Path Planning Algorithms
- Optical Imaging and Spectroscopy Techniques
- Ultrasound and Hyperthermia Applications
- Reinforcement Learning in Robotics
- Anatomy and Medical Technology
- Medical Image Segmentation Techniques
- Generative Adversarial Networks and Image Synthesis
- Aortic Disease and Treatment Approaches
Stanford University
2023-2024
Technical University of Munich
2018-2023
Acquiring good image quality is one of the main challenges for fully-automatic robot-assisted ultrasound systems (RUSS). The presented method aims at overcoming this challenge orthopaedic applications by optimizing orientation robotic (US) probe, i.e. aligning central axis US probe to tissue's surface normal point contact in order improve sound propagation within tissue. We first optimize in-plane analyzing confidence map image. then carry out a fan motion and analyze resulting forces...
In this paper we introduce the first reinforcement learning (RL) based robotic navigation method which utilizes ultrasound (US) images as an input. Our approach combines state-of-the-art RL techniques, specifically deep Q-networks (DQN) with memory buffers and a binary classifier for deciding when to terminate task.Our is trained evaluated on in-house collected data-set of 34 volunteers compared pure supervised (SL) it performs substantially better, highlights suitability US-guided...
Abstract Various morphological and functional parameters of peripheral nerves their vascular supply are indicative pathological changes due to injury or disease. Based on recent improvements in optoacoustic image quality, the ability multispectral tomography, investigate environment morphology is explored vivo a pilot study healthy volunteers tandem with ultrasound imaging (OPUS). The unique visualize vasa nervorum by observing intraneural vessels showcased for first time. In addition, it...
Abstract Purpose: The detection and treatment of abdominal aortic aneurysm (AAA), a vascular disorder with life-threatening consequences, is challenging due to its lack symptoms until it reaches critical size. Abdominal ultrasound (US) utilized for diagnosis; however, inherent low image quality reliance on operator expertise make computed tomography (CT) the preferred choice monitoring treatment. Moreover, CT datasets have been effectively used training deep neural networks aorta...
In medical imaging tasks, such as cardiac imaging, ultrasound acquisition time is crucial, however traditional high-quality beamforming techniques are computationally expensive and their performance hindered by sub-sampled data. To this end, we propose DeepFormer, a method to reconstruct high quality images in real-time on raw data performing an end-to-end deep learning-based reconstruction. Results vivo dataset of 19 participants show that DeepFormer offers promising advantages over...
Promising advances in flexible array technology present new opportunities for wearable ultrasound. However, arrays have non-rigid shapes and thus unknown element positions, posing a major challenge image reconstruction. We introduce robust method estimating the shape of ultrasound from data itself. By implementing delay-and-sum (DAS) beamformer an automatic differentiation framework, we show that phase error between translating subapertures can be used as loss function is backpropagated...
In this paper we propose a novel augmentation technique that improves not only the performance of deep neural networks on clean test data, but also significantly increases their robustness to random transformations, both affine and projective. Inspired by ManiFool, is performed line-search manifold-exploration method learns geometric transformations lead misclassification an image, while ensuring it remains same manifold as training data. This populates any dataset with images lie border...
Ultrasound imaging is caught between the quest for highest image quality, and necessity clinical usability. Our contribution two-fold: First, we propose a novel fully convolutional neural network ultrasound reconstruction. Second, custom loss function tailored to modality employed end-to-end training of network. We demonstrate that map time-delayed raw data minimum variance ground truth offers performance increases in environment. In doing so, path explored towards improved clinically viable...
In this paper, we propose a novel interpretation method tailored to histological Whole Slide Image (WSI) processing. A Deep Neural Network (DNN), inspired by Bag-of-Features models is equipped with Multiple Instance Learning (MIL) branch and trained weak supervision for WSI classification. MIL avoids label ambiguity enhances our model's expressive power without guiding its attention. We utilize fine-grained logit heatmap of the activations interpret decision-making process. The proposed...
Ultrasound imaging is becoming more prevalent in clinical practice and research. To counteract the drawbacks of high user-dependency difficult interpretability, ultrasound probes can be attached to robotic arms, enabling an increase accuracy repeatability. Currently, scans are mainly performed a perpendicular manner. However, these create shadows below high-attenuation structures like bones due acoustic shadowing, leading information loss scan. this improve compounding quality scans, we...
Ultrasound is widely used in medical diagnostics allowing for accessible and powerful imaging but suffers from resolution limitations due to diffraction the finite aperture of system, which restricts diagnostic use. The impulse function an ultrasound system called point spread (PSF), convolved with spatial distribution reflectors image formation process. Recovering high-resolution reflector distributions by removing distortions induced convolution process improves clarity detail....