arXiv subjects

Robin Krüger

Publications and source records attributed to Robin Krüger.

4 recordsLinked to original sources

arXiv2026

Breaking order: Talbot effect with spinodal architectures

The Talbot effect describes the emergence of periodic patterns in perturbed propagating wave fields. The effect is well studied for perturbations from structurally coherent optics such as diffraction gratings. The emergence of freeform and metaoptical designs raises the question of whether comparable behavior can also be observed from complex, non-periodic structures. Here we demonstrate that stochastic structures inspired by recent metamaterial designs, display a strong Talbot-like behavior. Re-emergence of projected wavefronts through stochastic spinodal architectures at distinct propagation distances are proven theoretically and experimentally in the visible and hard X-ray regimes. A direct application of this phenomenon is X-ray dark-field imaging for characterizing artificial and natural meso-structured materials. Our work shows that spinodal X-ray optics effectively bridge the gap between the two opposing approaches in dark-field X-ray imaging that advocate for either spatially fully coherent (i.e gratings) or incoherent (i.e diffusers) optics. This opens opportunities for exploring a new dimension in the implementation of X-ray imaging methods. Given the impact and universality of the classical Talbot effect, we expect our work to enable new opportunities for characterizing and manipulating matter.

physics.optics↗

arXiv2026

Benchmarking von ASR-Modellen im deutschen medizinischen Kontext: Eine Leistungsanalyse anhand von Anamnesegesprächen

Automatic Speech Recognition (ASR) offers significant potential to reduce the workload of medical personnel, for example, through the automation of documentation tasks. While numerous benchmarks exist for the English language, specific evaluations for the German-speaking medical context are still lacking, particularly regarding the inclusion of dialects. In this article, we present a curated dataset of simulated doctor-patient conversations and evaluate a total of 29 different ASR models. The test field encompasses both open-weights models from the Whisper, Voxtral, and Wav2Vec2 families as well as commercial state-of-the-art APIs (AssemblyAI, Deepgram). For evaluation, we utilize three different metrics (WER, CER, BLEU) and provide an outlook on qualitative semantic analysis. The results demonstrate significant performance differences between the models: while the best systems already achieve very good Word Error Rates (WER) of partly below 3%, the error rates of other models, especially concerning medical terminology or dialect-influenced variations, are considerably higher.

cs.CL↗

arXiv2025

Datenschutzkonformer LLM-Einsatz: Eine Open-Source-Referenzarchitektur

The development of Large Language Models (LLMs) has led to significant advancements in natural language processing and enabled numerous applications across various industries. However, many LLM-based solutions operate as open systems relying on cloud services, which pose risks to data confidentiality and security. To address these challenges, organizations require closed LLM systems that comply with data protection regulations while maintaining high performance. In this paper, we present a reference architecture for developing closed, LLM-based systems using open-source technologies. The architecture provides a flexible and transparent solution that meets strict data privacy and security requirements. We analyze the key challenges in implementing such systems, including computing resources, data management, scalability, and security risks. Additionally, we introduce an evaluation pipeline that enables a systematic assessment of system performance and compliance.

cs.CR↗

arXiv2024

Laboratory x-ray nano-computed tomography for biomedical research

High-resolution x-ray tomography is a common technique for biomedical research using synchrotron sources. With advancements in laboratory x-ray sources, an increasing number of experiments can be performed in the lab. In this paper, the design, implementation, and verification of a laboratory setup for x-ray nano-computed tomography is presented using a nano-focus x-ray source and high geometric magnification not requiring any optical elements. Comparing a scintillator-based detector to a photon counting detector shows a clear benefit of using photon counting detectors for these applications, where the flux of the x-ray source is limited and samples have low contrast. Sample contrast is enhanced using propagation-based phase contrast. The resolution of the system is verified using 2D resolution charts and using Fourier Ring Correlation on reconstructed CT slices. Evaluating noise and contrast highlights the benefits of photon counting detectors and the contrast improvement through phase contrast. The implemented setup is capable of reaching sub-micron resolution and satisfying contrast in biological samples, like paraffin embedded tissue.

physics.ins-det↗