68034 research outputs found
Sort by
Impact of Nitric Oxide on the Surface Properties of Selected Polymers
No full textThe change in the surface properties of polymer materials used in an extracorporeal membrane oxygenation (ECMO) device due to nitric oxide (NO) treatment was characterized by zeta-potential and dynamic contact-angle measurements. FTIR-ATR was used to determine the stability of these effects during liquid contact. Polymethyl pentene (PMP), methyl methacrylate acrylonitrile butadiene styrene (MABS), and polyurethane (PU) were investigated. The polymer materials were treated with NO (1000 ppm) for 17 h. The samples for FTIR-ATR measurements were submerged in water or physiological sodium chloride solution for 120 and 240 h after the end of the gas treatment. PMP showed no changes at all. MABS showed decreased contact-angles and increased contact-angle hysteresis. In contrast, PU showed decreased contact-angles and a shift in its zeta-potential curve, indicating a more hydrophilic and acidic surface. The FTIR-ATR measurements showed a slight decrease in the signal intensities after liquid contact. The results indicated an improvement in the liquid contact properties of MABS and the PU due to increased surface hydrophilicity caused mainly by the adsorbed nitric acid (HNO3) molecules formed by the NO treatment. The results presented in this paper point towards a simple and complication-free method of introducing NO into an ECMO circuit
Simultaneous determination of ceftazidime and avibactam in patients by isocratic ion-pair liquid chromatography with photometric detection
No full textA simple and fast HPLC-UV method is described for the simultaneous determination of total or free ceftazidime and avibactam in serum, which is suitable for therapeutic drug monitoring (TDM) or pharmacokinetic studies in man. Sufficient retention of the very polar avibactam was obtained by addition of tetrabutylammonium hydrogen sulfate (TBA) as ion pairing agent, thus allowing the determination of both drugs in serum by isocratic elution. Total concentrations were determined after protein precipitation with acetonitrile, free concentrations after ultrafiltration. Separation was performed using an XBridge BEH C18 column with a mobile phase consisting of 20 mM sodium phosphate buffer/acetonitrile 90:10 (v/v), pH 6.5, containing 5 mM TBA. The lower limit of quantification was 1 mg/L for ceftazidime and 0.5 mg/L for avibactam, respectively. The imprecision of the determination of total drug was <3 %, the accuracy between 99.0 % and 104 %. Determination of free drug in quality control samples resulted in unbound fractions of 97.9 ± 2.0 % for ceftazidime and 99.4 ± 3.2 % for avibactam
Data Collection in Cyber Exercises Through Monitoring Points: Observing, Steering, and Scoring
Full text linkCyber security exercises are an essential means to train people and increase their skill levels in IT operations, cyber incident response, and forensic investigations. Unfortunately, carrying out high-quality exercises requires tremendous human effort in planning, deploying, executing and evaluating well-planned cyber exercise scenarios. While planning a scenario is often only a one time effort, and deployment can be highly automatized today, their repeated execution and evaluation is a resource-intensive task. Usually human experts manually observe the participants to recognize any difficulties in carrying out the exercise and to keep track of the participants’ progress. This is an essential prerequisite to not only support participants during the exercise, but also to drive the scenario further through timely injects, and provide feedback after the exercise. All this manual effort makes exercises a costly activity, reduces scalability and hinders their wide adoption. We argue that with automating observations, recognizing participant progress with only little to no human effort, and even steering the delivery of customized injects, cyber exercises could be carried out much more cost-effective. In this paper, we therefore introduce the concept of monitoring points which enable the scenario-dependent collection of technical data and the calculation of behavior and progress metrics to rate participants in exercises. This is the foundational basis for steering an exercise on the one side, and evaluation on the other side. We showcase our concept and implementation in course of a demonstrator consisting of a cyber exercise comprising 14 participants and discuss its applicability
Assessing performance and explainability of convolutional neural networks in brain state decoding
Full text linkHumans are capable of great feats in their daily lives. Our minds receive a constant stream of information that needs to be processed and acted upon to ensure our survival. This incoming information may be processed or represented by distinct patterns of brain activity, or, in other words, by distinct brain states. In the neuroimaging domain, such brain states have been defined as patterns of (whole-) brain activity. However, to this day many cognitive tasks are still under investigation to figure out which brain areas are involved in their processing. Over the last decades, different approaches to reveal such brain states and their associated cognitive task have been proposed. The most commonly used approach has been forward inference. Here, researchers performed studies in which they could conclude that a stimulus (e.g., a visual stimulus) might have activated a specific brain area. Such analyses have commonly been done in a so-called massive univariate- or encoding analysis.
In the early 2000s a different approach, often called brain reading, decoding, or reverse inference, was introduced. In this approach researchers used, for example, machine learning (ML) algorithms to learn to label patterns of brain activity in regions of interests (ROIs) spanning an entire section of the brain or in so-called small spherical searchlights. With such ML algorithms researchers have been able to discriminate between patterns of brain activity which were associated with a given stimulus. A seminal study showed that with such approaches, differences in patterns of brain activity in the same brain areas were indicative for a given stimulus, thus suggesting that representations of stimuli are distributed but also overlapping in human cortex.
Now, with increased computational power, more complex types of ML algorithms can be used to identify brain states; specifically deep learning algorithms, e.g., so-called convolutional neural networks (CNNs). While they have been shown to outperform standard machine learning algorithms, they are considered a black box. That is, we know what the input and the predicted output is, but we do not know based on what and how a deep learning algorithm made its prediction.
However, knowing which input features were most important for a deep learning algorithm is of utmost importance for cognitive neuroscience. Only if we know whether a deep learning algorithm assigned high relevance to anatomically and functionally meaningful
1
Abstract
features (i.e., brain areas) can we make inference about which brain area or potential process were involved in a cognitive task.
A relatively new emerging field of explainable artificial intelligence (XAI) is trying to elucidate exactly that: which features of the input were important for an algorithm’s prediction? To this end, I developed a custom CNN architecture capable to discriminate between many different cognitive tasks as measured by functional magnetic resonance imaging (fMRI). I trained this CNN architecture using publicly available data from the Human Connectome Project (HCP) and compared its performance with a standard ML algorithm, a support vector machine (SVM).
This initial test was to investigate whether the CNN architecture indeed can discriminate between patterns of brain activity of different cognitive tasks. However, our main goal was to investigate whether the CNN has learned to discriminate these tasks based on anatomically and functionally meaningful features. Therefore, I used algorithms from the XAI domain, specifically the Guided Backpropagation- (GBP) and layer-wise relevance propagation (LRP) algorithms, to reveal features our CNN deemed as important for its prediction.
I found that our CNN architecture as well as the SVM can reliably discriminate between whole-brain patterns of fMRI data. It appeared that the SVM slightly outperforms the CNN architecture. However, the LRP algorithm revealed that, in some cognitive tasks, the CNN architecture used more anatomically and functionally sensible input features for its prediction compared to those used by the SVM or as identified by a univariate analysis.
Additional analyses revealed that so-called transfer learning allows the custom CNN architecture to already perform well with small sample sizes. This finding is important for neuroimagers since large datasets are commonly expensive and time consuming to acquire. Hence the transfer learning approach demonstrated that small sizes were feasible to use (at least in the HCP dataset).
Lastly, I demonstrate in the discussion section that the custom CNN architecture can decode brain states in real-time. This knowledge opens further potential experiments with therapeutic indication in patient populations with, e.g., psychiatric diseases
Breast cancer scoring based on a multiplexed profiling of soluble and cell-associated (immune) markers facilitates the prediction of pembrolizumab therapy
Full text linkBackground:
The immune checkpoint targeting is nowadays an integral part of cancer therapies. However, only a minority of patients experience long-term benefits. Thus, the identification of predictive biomarkers contributing to therapy response is urgently needed.
Methods:
Here, we analyzed different immune and tumor specific expression and secretion profiles in the peripheral blood and tumor samples of 50 breast cancer patients by multicolor flow cytometry and bead-based immunoassays at the time of diagnosis. Due to individual phenotype variations, we quantitatively scored 25 expressed and secreted immune-associated (e.g., LAG-3, PD-1, TIM-3, CD27) and tumor relevant markers (e.g., PD-L1, CD44, MHC-I, MHC-II) in immune checkpoint-treated triple negative breast cancer patients based on the current literature. The calculated score divided the patients into individuals with predicted pCR (total score of > 0) or predicted residual disease (total score of ≤ 0). At the end of the neoadjuvant therapy, the truly achieved pathological complete response (pCR; end of observation) was determined.
Results:
The calculated score was 79% in accordance with the achieved pCR at the time of surgery. Moreover, the sensitivity was 83.3%, the specificity 76.9%, the positive predictive value 62.5%, and the negative predictive value 90.9%. In addition, we identified a correlation of PD-1 and LAG-3 expression between tumor-associated and peripheral immune cells, which was independent of the subtype. Overall, PD-1 was the most frequently expressed checkpoint. However, in a number of patient-derived tumors, additional checkpoints as LAG-3 and TIM-3 were substantially (co-)expressed, which potentially compromises anti-PD-(L)1 mono-therapy.
Conclusions:
This study represents a proof-of-principle to identify potential checkpoint therapy responders in advance at the time of diagnosis. The work was based on a scoring derived from a multiplexed marker profiling. However, larger patient cohorts need to be prospectively evaluated for further validation
Fracture-Related Infection of the Proximal Femur – Diagnostics and Treatment
Full text linkPurpose: With the aging population and rising life expectancy the incidence of trauma-related injuries, particularly proximal femur fractures, is expected to increase. Complications such as fracture-related infections (FRI) significantly impede the healing process and pose substantial risks to patients. Despite advancements in understanding, diagnosing, and treating FRI, challenges
persist in achieving optimal outcomes. This review addresses the significance of FRI following proximal femur fractures, emphasizing diagnostic methodologies and therapeutic modalities to enhance clinical care. Findings: Notably, a consensus definition for FRI has been established, providing clarity for accurate diagnosis.Diagnostic criteria encompass confirmatory and
suggestive elements, facilitating precise identification of FRI. Therapeutic strategies for FRI in proximal femur fractures include a spectrum of surgical and antimicrobial approaches. Surgical interventions, ranging from debridement with implant retention over implant removal/exchange to staged conversions to arthroplasty, are tailored based on fracture stability, individual patient factors, and infection characteristics. The intricate decision-making process is elucidated, highlighting the importance of individualized treatment plans and multidisciplinary collaboration. Antimicrobial therapy plays a pivotal role in FRI management, with empirical regiments targeting common pathogens and local delivery systems offering sustained antibiotic release. Microbiological analysis and collaboration with infectious disease specialists should guide antimicrobial treatment and ensure optimal therapy efficacy. Conclusion: Managing FRI following proximal femur fractures requires a tailored, multidisciplinary approach. Treatment strategies should be guided by diagnostic precision, patient-specific considerations, and collaboration among surgical, infectious disease, and clinical teams. Implementing comprehensive therapeutic approaches is essential for mitigating the impact of FRI and improving patient outcomes
Partielle Etablierung eines Blood Air Barrier Model zur Bestimmung der Diffusion anästhesiologisch relevanter Tracersubstanzen
Full text linkGrenzschichten spielen im Körper an diversen Orten eine herausragende Rolle.
Besonders auch die Blut-Gas-Grenzschicht in der Lunge beeinflusst maßgeblich die Diffusion von überlebenswichtigen Molekülen wie O2 und CO2. Dies wird besonders bei pathophysiologischen Prozessen wie einem ARDS deutlich. Methoden zur experimentellen Untersuchung der Gasdiffusion an diesen Grenzschichten fehlen bisher. Das Ziel dieser Arbeit war eine Teiletablierung eines Blood Air Barrier Model und die Identifikation erster geeigneter Tracersubstanzen. Im Rahmen einer Methodenentwicklung wurde basierend auf einem Membranslide, Endothelzellen und einem Pumpensystem ein experimentelles Modell für den endothelseitigen Anteil der pulmonalen Diffusionsstrecke entwickelt. Mittels Gaschromatographie wurden Zeit-Werte-Paare für die Diffusion der Substanzen Ether und Sevofluran erhoben. Die Komplexität aus einer Zellkultur unter Perfusion in Kombination mit einem Membranslide stellte sich dabei als sehr herausfordernd dar. Über einen Zeitraum von 40 Minuten wurde die Tracerdiffusion durch einen Monolayer
aus Endothelzellen gemessen. Dabei war ein An- und Abfluten entsprechend einem 2-Kompartimente-Modell zu beobachten. Mit den erhobenen Daten konnte kein statistisch signifikanter Einfluss der Endothelzellschicht auf die Diffusion der Substanzen nachgewiesen werden. Es ist erstmals gelungen diese komplexe Methode zu etablieren. Zur weiteren Vervollständigung des Modells bedarf es einer zweiten Zellschicht. Es besteht Grund zur Annahme, dass mit weiterer Standardisierung der Methode auch eine erhöhte Sensitivität für Änderungen der Diffusion durch die Endothelzellschicht erreicht werden kann
Functional analysis of aberrant DNA methylation patterns during metastatic breast cancer progression
Full text linkAberrant DNA methylation in primary breast cancer tumors plays an important role in gene regulation and transcriptome diversity and influences metastatic behavior. Despite multiple hypotheses on how malignant cells spread and become metastatic, there is a lack of comprehensive analysis of how the methylome contributes to the process of dissemination. Previous studies have shown that metastatic cells accumulate hypermethylation at promoters and demethylation events in intra- and intergenic regions. These epigenetic changes contribute to increased chromosomal instability and the downregulation of tumor suppressor genes. Furthermore, analysis conducted by our laboratory has revealed the presence of epigenetic modifications in metastasis, indicating distinct recurrent patterns of promoter methylation and demethylation. However, it is unclear whether these epigenetic changes arise as side effects of cell aging and increased proliferation or through the acquisition of epigenetic alterations by single cells in early primary tumor lesions, which are then selected to become metastatic. To address this complex issue, a novel approach to analyze diverse methylation patterns throughout primary tumor progression was developed, which is the topic of this thesis. Specifically, a barcoded bisulfite amplicon deep sequencing (BBA-seq) technique was used to characterize the methylome on a single allelic level within breast cancer in the Her2-driven Balb-NeuT mouse model. Computational processing, clustering, and precise representation of the sequenced reads according to their methylation composition led to a clear resolution of the methylome in seven primary tumors and their corresponding metastases within 9 DNA regions. These regions include demethylation, hypermethylation, and intermediate conditions of DNA methylation. The findings revealed diverse methylation patterns, ranging from a constant increase in DNA methylation to the identification of sub-clones with an already high methylation character within primary tumors. These observations highlight the complexity and variability of methylation patterns during tumor progression. The results suggest that further investigation into the role of these sub-clones is critical in understanding how epigenetic changes drive metastasis. In addition to identifying and characterizing DNA methylation patterns, a comprehensive protocol was established to alter the DNA methylome via a gene knockdown of the de-novo DNA (cytosine-5)-methyltransferase 3a using a CRISPR/Cas9 nickase approach in hard-to-transfect primary mouse mammary gland cells. Overall, this thesis comprehensively analyzes DNA methylome remodeling during primary tumor progression and metastasis in breast cancer. The novel methodology developed in this study may also be applied to investigate other research questions related to DNA methylation in cancer biology and beyond
Die ewig veränderliche Zukunft des Handels
No full textLiebe Standpunkt-Leserinnen und -Leser,
schon lange gilt für den Einzelhandel, dass das Flächenwachstum und die Sortimentsvielfalt nicht mehr Wachstumsversprechen einlösen können. Stattdessen wird der Dreiklang aus Service, Erlebnis und Gastronomie empfohlen. Dies ist weiterhin richtig, doch es verkennt zwei weitere Chancen: Zusammenstellen von Ladenportfolios, um Wegekosten zu sparen, und das Kuratieren von Läden. Dort gewinnt eben nicht Größe und auch nicht Vielfalt, sondern das Einsparen von Transaktionskosten durch optimal reduziertes Angebot. Dies erläutere ich in dem neuen IREBS Standpunkt
Efficient detection of 1H, 15N correlations in hydrogen bonded low molecular catalyst–substrate intermediates without selective 15N-labelling
Full text linkTo date, SOFAST approaches have generally been limited to biomolecules. We present the applicability of SOFAST-HMQC techniques to small molecules in the slow-tumbling regime offering a time-efficient characterization of catalyst substrate hydrogen bonds with nitrogen at natural abundance. This extends NMR access to a broader range of catalyst substrate combinations