A5003925489- Tactile and Sensory Interactions
- Advanced Sensor and Energy Harvesting Materials
- Image Processing and 3D Reconstruction
- Interactive and Immersive Displays
- 3D Shape Modeling and Analysis
- Acute Ischemic Stroke Management
- Multimodal Machine Learning Applications
- Medical Imaging and Analysis
- Soft Robotics and Applications
- 3D Surveying and Cultural Heritage
- Transportation and Mobility Innovations
- Healthcare Operations and Scheduling Optimization
- Augmented Reality Applications
- Photoacoustic and Ultrasonic Imaging
- Advanced Neural Network Applications
- Generative Adversarial Networks and Image Synthesis
- Optical measurement and interference techniques
- Medical Image Segmentation Techniques
- Stroke Rehabilitation and Recovery
- Cerebrovascular and Carotid Artery Diseases
- Advanced Vision and Imaging
- Stochastic Gradient Optimization Techniques
- Explainable Artificial Intelligence (XAI)
University of Liverpool
2022-2024
Transferring optical tactile skills learned from simulated environments to the real world benefits many robotic applications, which can reduce cost of data collection. However, models purely trained on are often difficult generalize well unseen due domain gap between faultless training images and testing with unpredictable manufacturing defects or natural wear. In this paper, we propose an Adaptively Correlation-attentive Task-related Network (ACTNet) for image transfer, a novel unsupervised...
Endovascular navigation is a crucial aspect of minimally invasive procedures, where precise control curvilinear instruments like guidewires critical for successful interventions. A key challenge in this task accurately predicting the evolving shape guidewire as it navigates through vasculature, which presents complex deformations due to interactions with vessel walls. Traditional segmentation methods often fail provide accurate real-time predictions, limiting their effectiveness highly...
In endovascular surgery, the precise identification of catheters and guidewires in X-ray images is essential for reducing intervention risks. However, accurately segmenting catheter guidewire structures challenging due to limited availability labeled data. Foundation models offer a promising solution by enabling collection similar domain data train whose weights can be fine-tuned downstream tasks. Nonetheless, large-scale training constrained necessity maintaining patient privacy. This paper...
Recently simulation methods have been developed for optical tactile sensors to enable the Sim2Real learning, i.e., first training models in before deploying them on a real robot. However, some artefacts objects are unpredictable, such as imperfections caused by fabrication processes, or scratches natural wear and tear, thus cannot be represented simulation, resulting significant gap between simulated images. To address this gap, we propose novel texture generation network map images into...
Endovascular robots have been actively developed in both academia and industry. However, progress toward autonomous catheterization is often hampered by the widespread use of closed-source simulators physical phantoms. Additionally, acquisition large-scale datasets for training machine learning algorithms with endovascular usually infeasible due to expensive medical procedures. In this chapter, we introduce CathSim, first open-source simulator intervention address these limitations. CathSim...
Real-time visual feedback from catheterization analysis is crucial for enhancing surgical safety and efficiency during endovascular interventions. However, existing datasets are often limited to specific tasks, small scale, lack the comprehensive annotations necessary broader intervention understanding. To tackle these limitations, we introduce CathAction, a large-scale dataset Our CathAction encompasses approximately 500,000 annotated frames action understanding collision detection, 25,000...
Endovascular surgical tool reconstruction represents an important factor in advancing endovascular navigation, which is step surgery. However, the lack of publicly available datasets significantly restricts development and validation novel machine learning approaches. Moreover, due to need for specialized equipment such as biplanar scanners, most previous research employs monoplanar fluoroscopic technologies, hence only capturing data from a single view limiting accuracy. To bridge this gap,...
Endovascular intervention training is increasingly being conducted in virtual simulators. However, transferring the experience from endovascular simulators to real world remains an open problem. The key challenge environments are usually not realistically simulated, especially simulation images. In this paper, we propose a new method translate images simulator X-ray Previous image-to-image translation methods often focus on visual effects and neglect structure information, which critical for...
Endovascular navigation, essential for diagnosing and treating endovascular diseases, predominantly hinges on fluoroscopic images due to the constraints in sensory feedback. Current shape reconstruction techniques intervention often rely either a priori information or specialized equipment, potentially subjecting patients heightened radiation exposure. While deep learning holds potential, it typically demands extensive data. In this paper, we propose new method reconstruct 3D guidewire by...
Autonomous robots in endovascular operations have the potential to navigate circulatory systems safely and reliably while decreasing susceptibility human errors. However, there are numerous challenges involved with process of training such robots, as long duration safety issues arising from interaction between catheter aorta. Recently, simulators been employed for medical but generally do not conform autonomous catheterization. Furthermore, most current closed-source, which hinders...
Recently simulation methods have been developed for optical tactile sensors to enable the Sim2Real learning, i.e., firstly training models in before deploying them on real robot. However, some artefacts objects are unpredictable, such as imperfections caused by fabrication processes, or scratches natural wear and tear, thus cannot be represented simulation, resulting a significant gap between simulated images. To address this gap, we propose novel texture generation network that maps images...