Flensburg University of Applied Sciences

Muffels I.J., Waterham H.R., D’Alessandro G., Zagnoli-Vieira G., Sacher M., Lefeber D.J., Van der Vinne C., Roifman C.M., Gassen K.L., Rehmann H., Van Haaften-Visser D.Y., Nieuwenhuis E.S., Jackson S.P., Fuchs S.A., Wijk F., et. al.

Deciphering variants of uncertain significance (VUS) represents a major diagnostic challenge, partially due to the lack of easy-to-use and versatile cellular readouts that aid the interpretation of pathogenicity and pathophysiology. To address this challenge, we propose a high-throughput screening of cellular functionality through an imaging flow cytometry (IFC)-based platform. Six assays to evaluate autophagic-, lysosomal-, Golgi- health, mitochondrial function, ER stress, and NF-κβ activity were developed in fibroblasts. Assay sensitivity was verified with compounds (N = 5) and positive control patients (N = 6). Eight healthy controls and 20 individuals with VUS were screened. All molecular compounds and positive controls showed significant changes on their cognate assays, confirming assay sensitivity. Simultaneous screening of positive control patients on all six assays revealed distinct phenotypic profiles. In addition, individuals with VUS(es) in well-known disease genes showed distinct – but similar—phenotypic profiles compared to patients with pathogenic variants in the same gene.. For all individuals with VUSes in Genes of Uncertain Significance (GUS), we found one or more of six assays were significantly altered. Broadening the screening to an untargeted approach led to the identification of two clusters that allowed for the recognition of altered cell cycle dynamics and DNA damage repair defects. Experimental follow-up of the ‘DNA damage repair defect cluster’ led to the discovery of highly specific defects in top2cc release from double-strand DNA breaks in one of these individuals, harboring a VUS in the RAD54L2 gene. Our high-throughput IFC-based platform simplifies the process of identifying VUS pathogenicity through six assays and allows for the recognition of useful pathophysiological markers that structure follow-up experiments, thereby representing a novel valuable tool for precise functional diagnostics in genomics.

Reese L., Rettig A., Jauch C., Domin R.J., Karshüning T.

Due to the energy transition, the future electric power system will face further challenges that affect the functionality of the electricity grid and therefore the security of supply. For this reason, this article examines the future frequency stabilisation in a 100% renewable electric power system. A focus is set on the provision of inertia and frequency containment reserve. Today, the frequency stabilisation in most power systems is based on synchronous generators. By using grid-forming frequency converters, a large potential of alternative frequency stabilisation reserves can be tapped. Consequently, frequency stabilisation is not a problem of existing capacities but whether and how these are utilised. Therefore, in this paper, a collaborative approach to realise frequency stabilisation is proposed. By distributing the required inertia and frequency containment reserve across all technologies that are able to provide it, the relative contribution of each individual provider is low. To cover the need for frequency containment reserve, each capable technology would have to provide less than 1% of its rated power. The inertia demand can be covered by the available capacities at a coverage ratio of 171% (excluding wind power) to 217% (all capacities). As a result, it is proposed that provision of frequency stabilisation is made mandatory for all capable technologies. The joint provision distributes the burden of frequency stabilisation across many participants and hence increases redundancy. It ensures the stability of future electricity grids, and at the same time, it reduces the technological and economic effort. The findings are presented for the example of the German electricity grid.

Ganz J., Ammeling J., Jabari S., Breininger K., Aubreville M.

In numerous studies, deep learning algorithms have proven their potential for the analysis of histopathology images, for example, for revealing the subtypes of tumors or the primary origin of metastases. These models require large datasets for training, which must be anonymized to prevent possible patient identity leaks. This study demonstrates that even relatively simple deep learning algorithms can re-identify patients in large histopathology datasets with substantial accuracy. In addition, we compared a comprehensive set of state-of-the-art whole slide image classifiers and feature extractors for the given task. We evaluated our algorithms on two TCIA datasets including lung squamous cell carcinoma (LSCC) and lung adenocarcinoma (LUAD). We also demonstrate the algorithm's performance on an in-house dataset of meningioma tissue. We predicted the source patient of a slide with F

Kok G., Schene I.F., Ilcken E.F., Alcaraz P.S., Mendes M., Smith D.E., Salomons G., Shehata S., Jans J.J., Maroofian R., Hoek T.A., van Es R.M., Rehmann H., Nieuwenhuis E.S., Vos H.R., et. al.

Abstract Aminoacyl-tRNA synthetases (ARSs) couple tRNAs with their corresponding amino acids. While ARSs can bind structurally similar amino acids, extreme specificity is ensured by subsequent editing activity. Yet, we found that upon isoleucine (I) restriction, healthy fibroblasts consistently incorporated valine (V) into proteins at isoleucine codons, resulting from misacylation of tRNAIle with valine by wildtype IARS1. Using a dual-fluorescent reporter of translation, we found that valine supplementation could fully compensate for isoleucine depletion and restore translation to normal levels in healthy, but not IARS1 deficient cells. Similarly, the antiproliferative effects of isoleucine deprivation could be fully restored by valine supplementation in healthy, but not IARS1 deficient cells. This indicates I > V substitutions help prevent translational termination and maintain cellular function in human primary cells during isoleucine deprivation. We suggest that this is an example of a more general mechanism in mammalian cells to preserve translational speed at the cost of translational fidelity in response to (local) amino acid deficiencies.

Haghofer A., Parlak E., Bartel A., Donovan T.A., Assenmacher C., Bolfa P., Dark M.J., Fuchs-Baumgartinger A., Klang A., Jäger K., Klopfleisch R., Merz S., Richter B., Schulman F.Y., Janout H., et. al.

Variation in nuclear size and shape is an important criterion of malignancy for many tumor types; however, categorical estimates by pathologists have poor reproducibility. Measurements of nuclear characteristics can improve reproducibility, but current manual methods are time-consuming. The aim of this study was to explore the limitations of estimates and develop alternative morphometric solutions for canine cutaneous mast cell tumors (ccMCTs). We assessed the following nuclear evaluation methods for accuracy, reproducibility, and prognostic utility: (1) anisokaryosis estimates by 11 pathologists; (2) gold standard manual morphometry of at least 100 nuclei; (3) practicable manual morphometry with stratified sampling of 12 nuclei by 9 pathologists; and (4) automated morphometry using deep learning–based segmentation. The study included 96 ccMCTs with available outcome information. Inter-rater reproducibility of anisokaryosis estimates was low (k = 0.226), whereas it was good (intraclass correlation = 0.654) for practicable morphometry of the standard deviation (SD) of nuclear size. As compared with gold standard manual morphometry (area under the ROC curve [AUC] = 0.839, 95% confidence interval [CI] = 0.701–0.977), the prognostic value (tumor-specific survival) of SDs of nuclear area for practicable manual morphometry and automated morphometry were high with an AUC of 0.868 (95% CI = 0.737–0.991) and 0.943 (95% CI = 0.889–0.996), respectively. This study supports the use of manual morphometry with stratified sampling of 12 nuclei and algorithmic morphometry to overcome the poor reproducibility of estimates. Further studies are needed to validate our findings, determine inter-algorithmic reproducibility and algorithmic robustness, and explore tumor heterogeneity of nuclear features in entire tumor sections.

Ganz J., Marzahl C., Ammeling J., Rosbach E., Richter B., Puget C., Denk D., Demeter E.A., Tăbăran F.A., Wasinger G., Lipnik K., Tecilla M., Valentine M.J., Dark M.J., Abele N., et. al.

AbstractThe count of mitotic figures (MFs) observed in hematoxylin and eosin (H&E)-stained slides is an important prognostic marker, as it is a measure for tumor cell proliferation. However, the identification of MFs has a known low inter-rater agreement. In a computer-aided setting, deep learning algorithms can help to mitigate this, but they require large amounts of annotated data for training and validation. Furthermore, label noise introduced during the annotation process may impede the algorithms’ performance. Unlike H&E, where identification of MFs is based mainly on morphological features, the mitosis-specific antibody phospho-histone H3 (PHH3) specifically highlights MFs. Counting MFs on slides stained against PHH3 leads to higher agreement among raters and has therefore recently been used as a ground truth for the annotation of MFs in H&E. However, as PHH3 facilitates the recognition of cells indistinguishable from H&E staining alone, the use of this ground truth could potentially introduce an interpretation shift and even label noise into the H&E-related dataset, impacting model performance. This study analyzes the impact of PHH3-assisted MF annotation on inter-rater reliability and object level agreement through an extensive multi-rater experiment. Subsequently, MF detectors, including a novel dual-stain detector, were evaluated on the resulting datasets to investigate the influence of PHH3-assisted labeling on the models’ performance. We found that the annotators’ object-level agreement significantly increased when using PHH3-assisted labeling (F1: 0.53 to 0.74). However, this enhancement in label consistency did not translate to improved performance for H&E-based detectors, neither during the training phase nor the evaluation phase. Conversely, the dual-stain detector was able to benefit from the higher consistency. This reveals an information mismatch between the H&E and PHH3-stained images as the cause of this effect, which renders PHH3-assisted annotations not well-aligned for use with H&E-based detectors. Based on our findings, we propose an improved PHH3-assisted labeling procedure.

Qiu J., Aubreville M., Wilm F., Öttl M., Utz J., Schlereth M., Breininger K.

Acquiring annotations for whole slide images (WSIs)-based deep learning tasks, such as creating tissue segmentation masks or detecting mitotic figures, is a laborious process due to the extensive image size and the significant manual work involved in the annotation. This paper focuses on identifying and annotating specific image regions that optimize model training, given a limited annotation budget. While random sampling helps capture data variance by collecting annotation regions throughout the WSI, insufficient data curation may result in an inadequate representation of minority classes. Recent studies proposed diversity sampling to select a set of regions that maximally represent unique characteristics of the WSIs. This is done by pretraining on unlabeled data through self-supervised learning and then clustering all regions in the latent space. However, establishing the optimal number of clusters can be difficult and not all clusters are task-relevant. This paper presents prototype sampling, a new method for annotation region selection. It discovers regions exhibiting typical characteristics of each task-specific class. The process entails recognizing class prototypes from extensive histopathology image-caption databases and detecting unlabeled image regions that resemble these prototypes. Our results show that prototype sampling is more effective than random and diversity sampling in identifying annotation regions with valuable training information, resulting in improved model performance in semantic segmentation and mitotic figure detection tasks. Code is available at https://github.com/DeepMicroscopy/Prototype-sampling .

Ding W., Xiang D., Long D., Luo Z., Li Y.

In order to meet the demands of weak-link identification of mechanical structures in high-precision equipment, this paper introduces the concepts and equations of modal momentum and modal momentum difference ratio (MMDR), as well as an identification method of weak links based on modal momentum. In this method, the modal data of the mechanical structure is acquired first through experimental modal analysis or finite element analysis. The modal momentum of each measured point is calculated based on the modal data and transformed to the global coordinate system. Then, the MMDRs between the measured point with the largest magnitude of modal momentum and its adjacent measured points are calculated. Weak links are judged if the MMDR exceeds a predefined threshold. Twelve different multi-degrees-of-freedom systems incorporating weak links were designed and then identified using the proposed method. The accuracy of identification is up to 100% for these systems. The method was applied to a double pendulum angle milling head of a five-axis numerically controlled machine tool, and the result reveals that the side plate lacks bending stiffness.

Aubreville M., Ganz J., Ammeling J., Rosbach E., Gehrke T., Scherzad A., Hackenberg S., Goncalves M.

Multidisciplinary tumor boards are meetings where a team of medical specialists, including medical oncologists, radiation oncologists, radiologists, surgeons, and pathologists, collaborate to determine the best treatment plan for cancer patients. While decision-making in this context is logistically and cost-intensive, it has a significant positive effect on overall cancer survival. We evaluated the quality and accuracy of predictions by several large language models for recommending procedures by a Head and Neck Oncology tumor board, which we adapted for the task using parameter-efficient fine-tuning or in-context learning. Records were divided into two sets: n=229 used for training and n=100 records for validation of our approaches. Randomized, blinded, manual human expert classification was used to evaluate the different models. Treatment line congruence varied depending on the model, reaching up to 86%, with medically justifiable recommendations up to 98%. Parameter-efficient fine-tuning yielded better outcomes than in-context learning, and larger/commercial models tend to perform better. Providing precise, medically justifiable procedural recommendations for complex oncology patients is feasible. Extending the data corpus to a larger patient cohort and incorporating the latest guidelines, assuming the model can handle sufficient context length, could result in more factual and guideline-aligned responses and is anticipated to enhance model performance. We, therefore, encourage further research in this direction to improve the efficacy and reliability of large language models as support in medical decision-making processes.

Büttner C., Esterl K., Cußmann I., Epia Realpe C.A., Amme J., Nadal A.

Germany must decarbonise all energy sectors to meet international and national climate goals. This task necessitates linking the electricity with the gas, heat and mobility sectors. On the one hand, sector coupling increases the demand for electrical energy and changes well-known demand patterns requiring updates to the grid infrastructure. On the other hand, the newly coupled sectors offer flexibility options to support the grid infrastructure and reduce expansion needs. This study employs a highly detailed model of the German transmission grid to analyse the impact of sector coupling comprising additional electricity demands and flexibility options on grid and storage expansion needs in the year 2035. The results demonstrate that utilising flexibility options can reduce system costs and lower CO2 emissions. The research adheres to open source and open data principles, with all data and tools being publicly accessible.

Ahn E., Kaltschmitt M., Langmaack T., Neuling U., Remmele E., Scherzinger M., Thuneke K., Voß S.

Die in bestimmten Pflanzenteilen, insbesondere Samen, enthaltenen Öle und / oder Fette können durch technische Verfahren mit einer sehr hohen Reinheit abgetrennt werden. Diese Pflanzenöle bzw. -fette sowie deren Mischungen können, falls sie bestimmten Mindestqualitätsanforderungen entsprechen, direkt als Kraftstoff in dafür spezifizierten selbstzündenden Motoren eingesetzt werden. In weit größerem Umfang aber werden Öle und Fette pflanzlichen oder tierischen Ursprungs in ihrem natürlichen molekularen Aufbau verändert, damit die Makromoleküle den Kohlenwasserstoffketten, wie sie Bestandteile des Dieselkraftstoffs sind, ähnlicher werden oder sogar entsprechen. Dadurch können diese Kraftstoffe konventionellem, fossilem Dieselkraftstoff beigemischt oder auch als Reinkraftstoff in Dieselmotoren, die nicht oder nur geringfügig modifiziert sind, verwendet werden. Im Folgenden wird zunächst auf die Herstellung von normgerechten Pflanzenölen als Reinkraftstoff eingegangen und anschließend werden Ansätze und Konzepte einer Weiterverarbeitung von Ölen und Fetten pflanzlichen oder tierischen Ursprungs im Hinblick auf die Einhaltung definierter Kraftstoffeigenschaften erläutert.

Fischer D., Vith W., Unger J.L.

Particle emissions from marine activities next to gaseous emissions have attracted increasing attention in recent years, whether in the form of black carbon for its contribution to global warming or as fine particulate matter posing a threat to human health. Coastal areas are particularly affected by this. Hence, there is a great need for shipping to explore alternative fuels that both reduce greenhouse gas emissions, as anticipated through IMO, and also have the potential to reduce particle emissions significantly. This paper presents a comparative study of the particulate emissions of two novel synthetic/biofuels (GTL and HVO), which might, in part, substitute traditionally used distillate liquid fuels (e.g., MDO). HFO particulate emissions, in combination with an EGCS, formed the baseline. The main emphasis was laid on particle concentration (PN) and particulate matter (PM) emissions, combining gravimetric and particle number measurements. Measurements were conducted on a 0.72 MW research engine at different loads (25%, 50%, and 75%). The results show that novel fuels produce slightly fewer emissions than diesel fuel. Results also exhibit a clear trend that particle formation decreases as engine load increases. The EGCS only moderately reduces particle emissions for all complaint fuels, which is related to the formation of very fine particles, especially at high engine loads.

Petersen K.

Rainer and Wohlin showed that case studies are not well understood by reviewers and authors and thus they say that a given research is a case study when it is not. Rainer and Wohlin proposed a smell indicator (inspired by code smells) to identify case studies based on the frequency of occurrences of words, which performed better than human classifiers. With the emergence of ChatGPT, we evaluate ChatGPT to assess its performance in accurately identifying case studies. We also reflect on the results' implications for mapping studies, specifically data extraction. We used ChatGPT with the model GPT-4 to identify case studies and compared the result with the smell indicator for precision, recall, and accuracy. GPT-4 and the smell indicator perform similarly, with GPT-4 performing slightly better in some instances and the smell indicator (SI) in others. The advantage of GPT-4 is that it is based on the definition of case studies and provides traceability on how it reaches its conclusions. As GPT-4 performed well on the task and provides traceability, we should use and, with that, evaluate it on data extraction tasks, supporting us as authors.

Zascerinska J., Scheepers J., Kühn M.

Problem-solving skills refer to the 21st century skills and Top 10 Skills of 2025. Therefore, TVET sector in South Africa has to pay particular attention to the development of TVET students’ problem solving skills. However, different stakeholders in TVET have different views on the development of TVET students’ problem-solving skills. The purpose of the paper is to model TVET students’ problem-solving skills on the multi-sided needs analysis of TVET students’ problem-solving skills. TVET students’ problem-solving skills are modelled on the following: a) implementation of theoretical analysis; b) online interviews with TVET lecturers; and c) 20 semi-structured interviews with the representatives of 18 companies (including German companies) carried out in South Africa in 2021. The multi-sided analysis of TVET students’ problem-solving skills shows that there is a discrepancy between the educational theory and educational practice aimed at fostering TVET students’ problem-solving skills.

Dyussembekova N., Schütt R., Leiße I., Ralfs B.

During the energy transition, new types of electrical equipment, especially power electronic devices, are proposed to increase the flexibility of electricity distribution grids. One type is the solid-state transformer (SST), which offers excellent possibilities to improve the voltage quality in electricity distribution grids and integrate hybrid AC/DC grids. This paper compares SST to conventional copper-based power transformers (CPT) with and without an on-load tap changer (OLTC) and with additional downstream converters. For this purpose, a corresponding electricity distribution grid is set up in the power system analysis tool DIgSILENT PowerFactory 2022. A DC generator like a photovoltaic system, a DC load like an electric vehicle fast charging station, and an AC load are connected. Based on load flow simulations, the four power transformers are compared concerning voltage stability during a generator-based and a load-based scenario. The results of load flow simulations show that SSTs are most valuable when additional generators and loads are to be connected to the infrastructure, which would overload the existing grid equipment. The efficiency of using SSTs also depends on the parameters of the electrical grid, especially the lengths of the low-voltage (LV) lines. In addition, a flowchart-based decision process is proposed to support the decision-making process for the appropriate power transformer from an electrical perspective. Beyond these electrical properties, an evaluation matrix lists other relevant criteria like characteristics of the installation site, noise level, expected lifetime, and economic criteria that must be considered.