A5103125177- Domain Adaptation and Few-Shot Learning
- Multimodal Machine Learning Applications
- Radiomics and Machine Learning in Medical Imaging
- Advanced Image and Video Retrieval Techniques
- COVID-19 diagnosis using AI
- Advanced Neural Network Applications
- AI in cancer detection
- Face recognition and analysis
- Geophysical Methods and Applications
- Neutropenia and Cancer Infections
- Human Pose and Action Recognition
- Biosimilars and Bioanalytical Methods
- Video Surveillance and Tracking Methods
- Blood disorders and treatments
- Medical Image Segmentation Techniques
- Generative Adversarial Networks and Image Synthesis
- Image Retrieval and Classification Techniques
- Heat shock proteins research
- Artificial Intelligence in Healthcare and Education
- RNA Research and Splicing
- Hepatocellular Carcinoma Treatment and Prognosis
- Cancer-related molecular mechanisms research
- Adversarial Robustness in Machine Learning
- Anomaly Detection Techniques and Applications
- Digital Imaging for Blood Diseases
SciencePharma (Poland)
2023-2024
University of Tübingen
1997-2024
TH Bingen University of Applied Sciences
2022-2023
Heidelberg University
2019-2022
Université de Montréal
2020
University of Toronto
2020
Fontbonne University
2020
Centre Universitaire de Mila
2020
Stony Brook Medicine
2020
Heidelberg University
2019-2020
Across the world’s coronavirus disease 2019 (COVID-19) hot spots, need to streamline patient diagnosis and management has become more pressing than ever. As one of main imaging tools, chest X-rays (CXRs) are common, fast, non-invasive, relatively cheap, potentially bedside monitor progression disease. This paper describes first public COVID-19 image data collection as well a preliminary exploration possible use cases for data. dataset currently contains hundreds frontal view is largest...
In this work, we report the set-up and results of Liver Tumor Segmentation Benchmark (LiTS), which was organized in conjunction with IEEE International Symposium on Biomedical Imaging (ISBI) 2017 Conferences Medical Image Computing Computer-Assisted Intervention (MICCAI) 2018. The image dataset is diverse contains primary secondary tumors varied sizes appearances various lesion-to-background levels (hyper-/hypo-dense), created collaboration seven hospitals research institutions. Seventy-five...
Being able to spot defective parts is a critical component in large-scale industrial manufacturing. A particular challenge that we address this work the cold-start problem: fit model using nominal (non-defective) example images only. While handcrafted solutions per class are possible, goal build systems well simultaneously on many different tasks automatically. The best peforming approaches combine embeddings from ImageNet models with an outlier detection model. In paper, extend line of and...
Introduction The need to streamline patient management for coronavirus disease-19 (COVID-19) has become more pressing than ever. Chest X-rays (CXRs) provide a non-invasive (potentially bedside) tool monitor the progression of disease. In this study, we present severity score prediction model COVID-19 pneumonia frontal chest X-ray images. Such can gauge lung infections (and in general) that be used escalation or de-escalation care as well monitoring treatment efficacy, especially ICU. Methods...
In this work, we report the set-up and results of Liver Tumor Segmentation Benchmark (LiTS), which was organized in conjunction with IEEE International Symposium on Biomedical Imaging (ISBI) 2017 Conferences Medical Image Computing Computer-Assisted Intervention (MICCAI) 2018. The image dataset is diverse contains primary secondary tumors varied sizes appearances various lesion-to-background levels (hyper-/hypo-dense), created collaboration seven hospitals research institutions. Seventy-five...
Metric learning seeks to embed images of objects such that class-defined relations are captured by the embedding space. However, variability in is not just due different depicted object classes, but also depends on other latent characteristics as viewpoint or illumination. In addition these structured properties, random noise further obstructs visual interest. The common approach metric enforce a representation invariant under all factors ones contrast, we propose explicitly learn shared and...
Deep Metric Learning (DML) is arguably one of the most influential lines research for learning visual similarities with many proposed approaches every year. Although field benefits from rapid progress, divergence in training protocols, architectures, and parameter choices make an unbiased comparison difficult. To provide a consistent reference point, we revisit widely used DML objective functions conduct study crucial as well commonly neglected mini-batch sampling process. Under comparison,...
Stress granules (SGs) are formed in the cytosol as an acute response to environmental cues and activation of integrated stress (ISR), a central signaling pathway controlling protein synthesis. Using chronic virus infection model, we previously uncovered unique temporal control ISR resulting recurrent phases SG assembly disassembly. Here, elucidate molecular network generating this fluctuating by integrating quantitative experiments with mathematical modeling find that operates stochastic...
To determine whether nailfold capillary images, acquired using video capillaroscopy, can provide diagnostic information about diabetes and its complications.Nailfold capillaroscopy was performed in 120 adult patients with without type 1 or 2 diabetes, cardiovascular disease. Nailfold images were analyzed convolutional neural networks, a deep learning technique. Cross-validation used to develop test the ability of models predict five5 prespecified states (diabetes, high glycosylated...
Proposed biosimilar natalizumab (biosim-NTZ) PB006 is the first monoclonal antibody therapy developed for multiple sclerosis (MS) treatment.To evaluate matching efficacy, safety, and immunogenicity between biosim-NTZ reference (ref-NTZ) in patients with relapsing-remitting MS (RRMS).The Antelope trial was a phase 3, parallel-group, randomized, active-controlled study, conducted October 2019 March 2021, last patient follow-up visit on August 23, 2021. The study took place 48 centers 7...
The visual classification performance of vision-language models such as CLIP has been shown to benefit from additional semantic knowledge large language (LLMs) GPT-3. In particular, averaging over LLM-generated class descriptors, e.g. "waffle, which a round shape", can notably improve generalization performance. this work, we critically study behavior and propose WaffleCLIP, framework for zero-shot simply replaces descriptors with random character word descriptors. Without querying external...
Learning visual similarity requires to learn relations, typically between triplets of images. Albeit triplet approaches being powerful, their computational complexity mostly limits training only a subset all possible triplets. Thus, sampling strategies that decide when use which sample during learning are crucial. Currently, the prominent paradigm fixed or curriculum predefined before starts. However, problem truly calls for process adjusts based on actual state representation training. We,...
Deep Metric Learning (DML) aims to learn representation spaces on which semantic relations can simply be expressed through predefined distance metrics. Best performing approaches commonly leverage class proxies as sample stand-ins for better convergence and generalization. However, these proxy-methods solely optimize sample-proxy distances. Given the inherent non-bijectiveness of used functions, this induce locally isotropic distributions, leading crucial context being missed due...
Deep Metric Learning (DML) proposes to learn metric spaces which encode semantic similarities as embedding space distances. These should be transferable classes beyond those seen during training. Commonly, DML methods task networks solve contrastive ranking tasks defined over binary class assignments. However, such approaches ignore higher-level relations between the actual classes. This causes learned incomplete context and misrepresent relation classes, impacting generalizability of space....
Across the world's coronavirus disease 2019 (COVID-19) hot spots, need to streamline patient diagnosis and management has become more pressing than ever. As one of main imaging tools, chest X-rays (CXRs) are common, fast, non-invasive, relatively cheap, potentially bedside monitor progression disease. This paper describes first public COVID-19 image data collection as well a preliminary exploration possible use cases for data. dataset currently contains hundreds frontal view is largest...
Learning the similarity between images constitutes foundation for numerous vision tasks. The common paradigm is discriminative metric learning, which seeks an embedding that separates different training classes. However, main challenge to learn a not only generalizes from novel, but related, test samples. It should also transfer object So what complementary information missed by paradigm? Besides finding characteristics <italic xmlns:mml="http://www.w3.org/1998/Math/MathML"...
At present, lesion segmentation is still performed manually (or semi-automatically) by medical experts. To facilitate this process, we contribute a fully-automatic pipeline. This work proposes method as part of the LiTS (Liver Tumor Segmentation Challenge) competition for ISBI 17 and MICCAI comparing methods automatics egmentation liver lesions in CT scans. By utilizing cascaded, densely connected 2D U-Nets Tversky-coefficient based loss function, our framework achieves very good shape...