A5072989453- Medical Image Segmentation Techniques
- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- Human Pose and Action Recognition
- Robot Manipulation and Learning
- Image Retrieval and Classification Techniques
- Robotic Mechanisms and Dynamics
- Ovarian cancer diagnosis and treatment
- Cancer-related molecular mechanisms research
- Medical Imaging and Analysis
- Brain Tumor Detection and Classification
- Animal Virus Infections Studies
- Image Processing and 3D Reconstruction
- Medical Imaging Techniques and Applications
Technical University of Munich
2022-2024
Medical image registration aims to identify the spatial deformation between images of same anatomical region and is fundamental image-based diagnostics therapy. To date, majority deep learning-based methods employ regularizers that enforce global smoothness, e.g., diffusion regularizer. However, such are not tailored data might be capable reflecting complex underlying deformation. In contrast, physics-inspired promote physically plausible deformations. One regularizer linear elastic...
Manipulability ellipsoids efficiently capture the human pose and reveal information about task at hand. Their use in task-dependent robot teaching - particularly their transfer from a teacher to learner can advance emulation of human-like motion. Although recent literature focus is shifted towards manipulability between two robots, adaptation capabilities other kinematic system date not addressed research still its infancy. This work presents novel domain method for another system. As...
General vision encoders like DINOv2 and SAM have recently transformed computer vision. Even though they are trained on natural images, such encoder models excelled in medical imaging, e.g., classification, segmentation, registration. However, no in-depth comparison of different state-of-the-art general for registration is available. In this work, we investigate how well features can be used the dissimilarity metrics image We explore two that were images as one was fine-tuned data. apply...
Physics-inspired regularization is desired for intra-patient image registration since it can effectively capture the biomechanical characteristics of anatomical structures. However, a major challenge lies in reliance on physical parameters: Parameter estimations vary widely across literature, and properties themselves are inherently subject-specific. In this work, we introduce novel data-driven method that leverages hypernetworks to learn tissue-dependent elasticity parameters an elastic...
Medical image registration aims at identifying the spatial deformation between images of same anatomical region and is fundamental to image-based diagnostics therapy. To date, majority deep learning-based methods employ regularizers that enforce global smoothness, e.g., diffusion regularizer. However, such are not tailored data might be capable reflecting complex underlying deformation. In contrast, physics-inspired promote physically plausible deformations. One regularizer linear elastic...
Manipulability ellipsoids efficiently capture the human pose and reveal information about task at hand. Their use in task-dependent robot teaching - particularly their transfer from a teacher to learner can advance emulation of human-like motion. Although recent literature focus is shifted towards manipulability between two robots, adaptation capabilities other kinematic system date not addressed research still its infancy. This work presents novel domain method for another system. As...