A5058612735- Acute Ischemic Stroke Management
- Cerebrovascular and Carotid Artery Diseases
- Radiomics and Machine Learning in Medical Imaging
- Traumatic Brain Injury and Neurovascular Disturbances
- Advanced MRI Techniques and Applications
- Retinal Imaging and Analysis
- Advanced Radiotherapy Techniques
- Machine Learning in Healthcare
- Artificial Intelligence in Healthcare and Education
- Explainable Artificial Intelligence (XAI)
- MRI in cancer diagnosis
- AI in cancer detection
- Stroke Rehabilitation and Recovery
- Generative Adversarial Networks and Image Synthesis
- Medical Image Segmentation Techniques
- Functional Brain Connectivity Studies
- EEG and Brain-Computer Interfaces
- Lung Cancer Diagnosis and Treatment
- Head and Neck Cancer Studies
- Autopsy Techniques and Outcomes
- Optical Imaging and Spectroscopy Techniques
- Radiation Dose and Imaging
- Visual perception and processing mechanisms
- Advanced Neural Network Applications
- Neural dynamics and brain function
Google (United States)
2023
Charité - Universitätsmedizin Berlin
2015-2022
Google (United Kingdom)
2020-2022
Berlin Institute of Health at Charité - Universitätsmedizin Berlin
2019
Humboldt-Universität zu Berlin
2019
Freie Universität Berlin
2019
Aarhus University
2017-2018
Max Planck Institute for Metabolism Research
2015-2017
Fraunhofer Institute for Digital Medicine
2015
Over half a million individuals are diagnosed with head and neck cancer each year globally. Radiotherapy is an important curative treatment for this disease, but it requires manual time to delineate radiosensitive organs at risk. This planning process can delay while also introducing interoperator variability, resulting in downstream radiation dose differences. Although auto-segmentation algorithms offer potentially time-saving solution, the challenges defining, quantifying, achieving expert...
Brain vessel status is a promising biomarker for better prevention and treatment in cerebrovascular disease. However, classic rule-based segmentation algorithms need to be hand-crafted are insufficiently validated. A specialized deep learning method-the U-net-is alternative. Using labeled data from 66 patients with disease, the U-net framework was optimized evaluated three metrics: Dice coefficient, 95% Hausdorff distance (95HD) average (AVD). The model performance compared traditional...
Over half a million individuals are diagnosed with head and neck cancer each year worldwide. Radiotherapy is an important curative treatment for this disease, but it requires manual time consuming delineation of radio-sensitive organs at risk (OARs). This planning process can delay treatment, while also introducing inter-operator variability resulting downstream radiation dose differences. While auto-segmentation algorithms offer potentially time-saving solution, the challenges in defining,...
State-of-the-art machine learning (ML) artificial intelligence methods are increasingly leveraged in clinical predictive modeling to provide decision support systems physicians. Modern ML approaches such as neural networks (ANNs) and tree boosting often perform better than more traditional like logistic regression. On the other hand, these modern yield a limited understanding of resulting predictions. However, medical domain, applied models is essential, particular, when informing support....
Explainability for artificial intelligence (AI) in medicine is a hotly debated topic. Our paper presents review of the key arguments favor and against explainability AI-powered Clinical Decision Support System (CDSS) applied to concrete use case, namely an CDSS currently used emergency call setting identify patients with life-threatening cardiac arrest. More specifically, we performed normative analysis using socio-technical scenarios provide nuanced account role CDSSs allowing abstractions...
Introduction: Arterial brain vessel assessment is crucial for the diagnostic process in patients with cerebrovascular disease. Non-invasive neuroimaging techniques, such as time-of-flight (TOF) magnetic resonance angiography (MRA) imaging are applied clinical routine to depict arteries. They are, however, only visually assessed. Fully automated segmentation integrated into could facilitate time-critical diagnosis of abnormalities and might identification valuable biomarkers events. In...
Stroke imaging is pivotal for diagnosis and stratification of patients with acute ischemic stroke to treatment. The potential combining multimodal information into reliable estimates outcome learning calls robust machine techniques high flexibility accuracy. We applied the novel extreme gradient boosting algorithm magnetic resonance imaging-based infarct prediction.In a retrospective analysis 195 stroke, fluid-attenuated inversion recovery, diffusion-weighted imaging, 10 perfusion parameters...
Deep learning requires large labeled datasets that are difficult to gather in medical imaging due data privacy issues and time-consuming manual labeling. Generative Adversarial Networks (GANs) can alleviate these challenges enabling synthesis of shareable data. While 2D GANs have been used generate images with their corresponding labels, they cannot capture the volumetric information 3D imaging. more suitable for this volumes but not labels. One reason might be synthesizing is challenging...
Organ-at-risk segmentation for head and neck cancer radiation therapy is a complex time-consuming process (requiring up to 42 individual structure, may delay start of treatment or even limit access function-preserving care. Feasibility using deep learning (DL) based autosegmentation model reduce contouring time without compromising contour accuracy assessed through blinded randomized trial oncologists (ROs) retrospective, de-identified patient data.
Reliable prediction of outcomes aneurysmal subarachnoid hemorrhage (aSAH) based on factors available at patient admission may support responsible allocation resources as well treatment decisions. Radiographic and clinical scoring systems help clinicians estimate disease severity, but their predictive value is limited, especially in devising strategies. In this study, we aimed to examine whether a machine learning (ML) approach using variables improve outcome aSAH compared established...
Background and Purpose— Dynamic susceptibility–weighted contrast–enhanced (DSC) magnetic resonance imaging (MRI) is used to identify the tissue-at-risk in acute stroke, but choice of optimal DSC postprocessing clinical setting remains a matter debate. Using 15O-water positron emission tomography (PET), we validated performance 2 common deconvolution methods for DSC-MRI. Methods— In (sub)acute stroke patients with consecutive MRI PET imaging, maps were calculated applying methods, standard...
Abstract Introduction Arterial brain vessel assessment is crucial for the diagnostic process in patients with cerebrovascular disease. Noninvasive neuroimaging techniques such as time-of-flight (TOF) magnetic resonance angiography (MRA) imaging are applied clinical routine to depict arteries. They are, however, only visually assessed. Fully automated segmentation integrated into could facilitate time-critical diagnosis of abnormalities and might identification valuable biomarkers events. In...
Abstract Background Cerebrovascular disease, in particular stroke, is a major public health challenge. An important biomarker cerebral hemodynamics. To measure and quantify hemodynamics, however, only invasive, potentially harmful or time-to-treatment prolonging methods are available. Results We present simulation-based approach which allows calculation of hemodynamics based on the patient-individual vessel configuration derived from structural imaging. For this, we implemented framework...
Brain arteries are routinely imaged in the clinical setting by various modalities, e.g., time-of-flight magnetic resonance angiography (TOF-MRA). These imaging techniques have great potential for diagnosis of cerebrovascular disease, disease progression, and response to treatment. Currently, however, only qualitative assessment is implemented applications, relying on visual inspection. While manual or semi-automated approaches quantification exist, such solutions impractical as they...
Abstract Background and Purpose Brain perfusion measurement in the subacute phase of stroke may support therapeutic decisions. We evaluated whether arterial spin labeling (ASL), a noninvasive imaging technique based on magnetic resonance (MRI), adds diagnostic prognostic benefit to diffusion‐weighted (DWI) stroke. Methods In single‐center study, patients with DWI lesion(s) middle cerebral artery (MCA) territory were included. Onset time was ≤7 days included ASL sequences. Qualitative...
Abstract Individualized treatment of acute stroke depends on the timely detection ischemia and potentially salvageable tissue in brain. Using functional MRI (fMRI), it is possible to characterize cerebral blood flow from blood‐oxygen‐level‐dependent (BOLD) signals without administration exogenous contrast agents. In this study, we applied spatial independent component analysis resting‐state fMRI data 37 patients scanned within 24 hr symptom onset, 17 whom received follow‐up scans next day....
ABSTRACT BACKGROUND AND PURPOSE In acute stroke, arterial‐input‐function (AIF) determination is essential for obtaining perfusion estimates with dynamic susceptibility‐weighted contrast‐enhanced magnetic resonance imaging (DSC‐MRI). Standard DSC‐MRI postprocessing applies single AIF selection, ie, global AIF. Physiological considerations, however, suggest that a multiple AIFs selection method would improve to detect penumbral flow. this study, we developed framework based on comparable and...
Identification of salvageable penumbra tissue by dynamic susceptibility contrast magnetic resonance imaging is a valuable tool for acute stroke patient stratification treatment. However, prior studies have not attempted to combine the different perfusion maps into predictive model. In this study, we established multiparametric model and cross-validated it using positron emission tomography detection penumbral flow.In retrospective analysis 17 subacute patients with consecutive H2O15 scans,...
Abstract Background State-of-the-art machine learning (ML) artificial intelligence methods are increasingly leveraged in clinical predictive modeling to provide decision support systems physicians. Modern ML approaches such as neural networks (ANNs) and tree boosting often perform better than more traditional like logistic regression. On the other hand, these modern yield a limited understanding of resulting predictions. However, medical domain, applied models is essential, particular, when...
Abstract Introduction Cerebrovascular disease is a major public health challenge. An important biomarker cerebral hemodynamics. To measure hemodynamics, however, only invasive, potentially harmful or time-to-treatment prolonging methods are available. We present simulation-based alternative which allows calculation of hemodynamics based on the individual vessel con figuration patient derived from structural imaging. Methods implemented framework allowing annotation extracted brain vessels...
Abstract Background and purpose Handling missing values is a prevalent challenge in the analysis of clinical data. The rise data-driven models demands an efficient use available Methods to impute are thus crucial. Here, we developed publicly framework test different imputation methods compared their impact typical stroke dataset as case. A based on 1000Plus study with 380 completed-entries patients was used. 13 common parameters including numerical categorical were selected. Missing...