Image-based recognition of parasitoid wasps using advanced neural networks.

Hymenoptera has some of the highest diversity and number of individuals among insects. Many of these species potentially play key roles as food sources, pest controllers and pollinators. However, little is known about the diversity and biology and ~80% of the species have not yet been described. Classical taxonomy based on morphology is a rather slow process but DNA barcoding has already brought considerable progress in identification. Innovative methods such as image-based identification and automation can further speed up the process. We present a proof of concept for image data recognition of a parasitic wasp family, the Diapriidae (Hymenoptera), obtained as part of the GBOL III project. These tiny (1.2-4.5mm) wasps were photographed and identified using DNA barcoding to provide a solid ground truth for training a neural network. Taxonomic identification was used down to the genus level. Subsequently, three different neural network architectures were trained, evaluated and optimised. As a result, 11 different genera of diaprids and one mixed group of 'other Hymenoptera' can be classified with an average accuracy of 96%. Additionally, the sex of the specimen can be classified automatically with an accuracy of >97%.

Surface-Patterned DNA Origami Rulers Reveal Nanoscale Distance Dependency of the Epidermal Growth Factor Receptor Activation.

The nanoscale arrangement of ligands can have a major effect on the activation of membrane receptor proteins and thus cellular communication mechanisms. Here we report on the technological development and use of tailored DNA origami-based molecular rulers to fabricate "Multiscale Origami Structures As Interface for Cells" (MOSAIC), to enable the systematic investigation of the effect of the nanoscale spacing of epidermal growth factor (EGF) ligands on the activation of the EGF receptor (EGFR). MOSAIC-based analyses revealed that EGF distances of about 30-40 nm led to the highest response in EGFR activation of adherent MCF7 and Hela cells. Our study emphasizes the significance of DNA-based platforms for the detailed investigation of the molecular mechanisms of cellular signaling cascades.

TRIM11 expression in non-small cell lung cancer is associated with poor prognosis.

BACKGROUND: Despite promising results of targeted therapy approaches, non-small cell lung cancer (NSCLC) remains the leading cause of cancer-related death. Tripartite motif containing 11 (TRIM11) is part of the TRIM family of proteins, playing crucial roles in tumor progression. TRIM11 serves as an oncogene in various cancer types and has been reported to be associated with a poor prognosis. In this study, we aimed to investigate the protein expression of TRIM11 in a large NSCLC cohort and to correlate its expression with comprehensive clinico-pathological data. METHODS: Immunohistochemical staining of TRIM11 was performed on a European cohort of NSCLC patients (n=275) including 224 adenocarcinomas and 51 squamous cell carcinomas. Protein expression was categorized according to staining intensity as absent, low, moderate and high. To dichotomize samples, absent and low expression was defined as weak and moderate and high expression was defined as high. Results were correlated with clinico-pathological data. RESULTS: TRIM11 was significantly more highly expressed in NSCLC than in normal lung tissue and significantly more highly expressed in squamous cell carcinomas than in adenocarcinomas. We found a significantly worse 5-year overall survival for patients who highly expressed TRIM11 in NSCLC. CONCLUSIONS: High TRIM11 expression is linked with a poor prognosis and might serve as a promising novel prognostic biomarker for NSCLC. Its assessment could be implemented in future routine diagnostic workup.

AI2Seg: A Method and Tool for AI-based Annotation Inspection of Biomedical Instance Segmentation Datasets.

In biomedical engineering, deep neural networks are commonly used for the diagnosis and assessment of diseases through the interpretation of medical images. The effectiveness of these networks relies heavily on the availability of annotated datasets for training. However, obtaining noise-free and consistent annotations from experts, such as pathologists, radiologists, and biologists, remains a significant challenge. One common task in clinical practice and biological imaging applications is instance segmentation. Though, there is currently a lack of methods and open-source tools for the automated inspection of biomedical instance segmentation datasets concerning noisy annotations. To address this issue, we propose a novel deep learning-based approach for inspecting noisy annotations and provide an accompanying software implementation, AI2Seg, to facilitate its use by domain experts. The performance of the proposed algorithm is demonstrated on the medical MoNuSeg dataset and the biological LIVECell dataset.

Repurposing FDA-Approved Drugs for Temozolomide-Resistant IDH1 Mutant Glioma Using High-Throughput Miniaturized Screening on Droplet Microarray Chip.

To address the challenge of drug resistance and limited treatment options for recurrent gliomas with IDH1 mutations, a highly miniaturized screening of 2208 FDA-approved drugs is conducted using a high-throughput droplet microarray (DMA) platform. Two patient-derived temozolomide-resistant tumorspheres harboring endogenous IDH1 mutations (IDH1mut ) are utilized. Screening identifies over 20 drugs, including verteporfin (VP), that significantly affected tumorsphere formation and viability. Proteomics analysis reveals that nuclear pore complex may be a potential VP target, suggesting a new mechanism of action independent of its known effects on YAP1. Knockdown experiments exclude YAP1 as a drug target in tumorspheres. Pathway analysis shows that NUP107 is a potential upstream regulator associated with VP response. Analysis of publicly available genomic datasets shows a significant correlation between high NUP107 expression and decreased survival in IDH1mut astrocytoma, suggesting NUP107 may be a potential biomarker for VP response. This study demonstrates a miniaturized approach for cost-effective drug repurposing using 3D glioma models and identifies nuclear pore complex as a potential target for drug development. The findings provide preclinical evidence to support in vivo and clinical studies of VP and other identified compounds to treat IDH1mut gliomas, which may ultimately improve clinical outcomes for patients with this challenging disease.

Self-Supervised Learning for Annotation Efficient Biomedical Image Segmentation.

OBJECTIVE: The scarcity of high-quality annotated data is omnipresent in machine learning. Especially in biomedical segmentation applications, experts need to spend a lot of their time into annotating due to the complexity. Hence, methods to reduce such efforts are desired. METHODS: Self-Supervised Learning (SSL) is an emerging field that increases performance when unannotated data is present. However, profound studies regarding segmentation tasks and small datasets are still absent. A comprehensive qualitative and quantitative evaluation is conducted, examining SSL's applicability with a focus on biomedical imaging. We consider various metrics and introduce multiple novel application-specific measures. All metrics and state-of-the-art methods are provided in a directly applicable software package (https://osf.io/gu2t8/). RESULTS: We show that SSL can lead to performance improvements of up to 10%, which is especially notable for methods designed for segmentation tasks. CONCLUSION: SSL is a sensible approach to data-efficient learning, especially for biomedical applications, where generating annotations requires much effort. Additionally, our extensive evaluation pipeline is vital since there are significant differences between the various approaches. SIGNIFICANCE: We provide biomedical practitioners with an overview of innovative data-efficient solutions and a novel toolbox for their own application of new approaches. Our pipeline for analyzing SSL methods is provided as a ready-to-use software package.

Synthesis of large scale 3D microscopic images of 3D cell cultures for training and benchmarking.

The analysis of 3D microscopic cell culture images plays a vital role in the development of new therapeutics. While 3D cell cultures offer a greater similarity to the human organism than adherent cell cultures, they introduce new challenges for automatic evaluation, like increased heterogeneity. Deep learning algorithms are able to outperform conventional analysis methods in such conditions but require a large amount of training data. Due to data size and complexity, the manual annotation of 3D images to generate large datasets is a nearly impossible task. We therefore propose a pipeline that combines conventional simulation methods with deep-learning-based optimization to generate large 3D synthetic images of 3D cell cultures where the labels are known by design. The hybrid procedure helps to keep the generated image structures consistent with the underlying labels. A new approach and an additional measure are introduced to model and evaluate the reduced brightness and quality in deeper image regions. Our analyses show that the deep learning optimization step consistently improves the quality of the generated images. We could also demonstrate that a deep learning segmentation model trained with our synthetic data outperforms a classical segmentation method on real image data. The presented synthesis method allows selecting a segmentation model most suitable for the user's data, providing an ideal basis for further data analysis.

Automated high-throughput image processing as part of the screening platform for personalized oncology.

Cancer is a devastating disease and the second leading cause of death worldwide. However, the development of resistance to current therapies is making cancer treatment more difficult. Combining the multi-omics data of individual tumors with information on their in-vitro Drug Sensitivity and Resistance Test (DSRT) can help to determine the appropriate therapy for each patient. Miniaturized high-throughput technologies, such as the droplet microarray, enable personalized oncology. We are developing a platform that incorporates DSRT profiling workflows from minute amounts of cellular material and reagents. Experimental results often rely on image-based readout techniques, where images are often constructed in grid-like structures with heterogeneous image processing targets. However, manual image analysis is time-consuming, not reproducible, and impossible for high-throughput experiments due to the amount of data generated. Therefore, automated image processing solutions are an essential component of a screening platform for personalized oncology. We present our comprehensive concept that considers assisted image annotation, algorithms for image processing of grid-like high-throughput experiments, and enhanced learning processes. In addition, the concept includes the deployment of processing pipelines. Details of the computation and implementation are presented. In particular, we outline solutions for linking automated image processing for personalized oncology with high-performance computing. Finally, we demonstrate the advantages of our proposal, using image data from heterogeneous practical experiments and challenges.

Automated Zebrafish Phenotype Pattern Recognition: 6 Years Ago, and Now.

The article assesses the developments in automated phenotype pattern recognition: Potential spikes in classification performance, even when facing the common small-scale biomedical data set, and as a reader, you will find out about changes in the development effort and complexity for researchers and practitioners. After reading, you will be aware of the benefits and unreasonable effectiveness and ease of use of an automated end-to-end deep learning pipeline for classification tasks of biomedical perception systems.

KaIDA: a modular tool for assisting image annotation in deep learning.

Deep learning models achieve high-quality results in image processing. However, to robustly optimize parameters of deep neural networks, large annotated datasets are needed. Image annotation is often performed manually by experts without a comprehensive tool for assistance which is time- consuming, burdensome, and not intuitive. Using the here presented modular Karlsruhe Image Data Annotation (KaIDA) tool, for the first time assisted annotation in various image processing tasks is possible to support users during this process. It aims to simplify annotation, increase user efficiency, enhance annotation quality, and provide additional useful annotation-related functionalities. KaIDA is available open-source at https://git.scc.kit.edu/sc1357/kaida.

Droplet Microarray Based Screening Identifies Proteins for Maintaining Pluripotency of hiPSCs.

Human induced pluripotent stem cells (hiPSCs) are crucial for disease modeling, drug discovery, and personalized medicine. Animal-derived materials hinderapplications of hiPSCs in medical fields. Thus, novel and well-defined substrate coatings capable of maintaining hiPSC pluripotency are important for advancing biomedical applications of hiPSCs. Here a miniaturized droplet microarray (DMA) platform to investigate 11 well-defined proteins, their 55 binary and 165 ternary combinations for their ability to maintainpluripotency of hiPSCs when applied as a surface coating, is used. Using this screening approach, ten protein group coatings are identified, which promote significantly higher NANOG expression of hiPSCs in comparison with Matrigel coating. With two of the identified coatings, long-term pluripotency maintenance of hiPSCs and subsequent differentiation into three germ layers are achieved. Compared with conventional high-throughput screening (HTS) in 96-well plates, the DMA platform uses only 83 µL of protein solution (0.83 µg total protein) and only ≈2.8 × 105 cells, decreasing the amount of proteins and cells ≈860 and 25-fold, respectively. The identified proteins will be essential for research and applications using hiPSCs, while the DMA platform demonstrates great potential for miniaturized HTS of scarce cells or expensive materials such as recombinant proteins.

Normics: Proteomic Normalization by Variance and Data-Inherent Correlation Structure.

Several algorithms for the normalization of proteomic data are currently available, each based on a priori assumptions. Among these is the extent to which differential expression (DE) can be present in the dataset. This factor is usually unknown in explorative biomarker screens. Simultaneously, the increasing depth of proteomic analyses often requires the selection of subsets with a high probability of being DE to obtain meaningful results in downstream bioinformatical analyses. Based on the relationship of technical variation and (true) biological DE of an unknown share of proteins, we propose the "Normics" algorithm: Proteins are ranked based on their expression level-corrected variance and the mean correlation with all other proteins. The latter serves as a novel indicator of the non-DE likelihood of a protein in a given dataset. Subsequent normalization is based on a subset of non-DE proteins only. No a priori information such as batch, clinical, or replicate group is necessary. Simulation data demonstrated robust and superior performance across a wide range of stochastically chosen parameters. Five publicly available spike-in and biologically variant datasets were reliably and quantitively accurately normalized by Normics with improved performance compared to standard variance stabilization as well as median, quantile, and LOESS normalizations. In complex biological datasets Normics correctly determined proteins as being DE that had been cross-validated by an independent transcriptome analysis of the same samples. In both complex datasets Normics identified the most DE proteins. We demonstrate that combining variance analysis and data-inherent correlation structure to identify non-DE proteins improves data normalization. Standard normalization algorithms can be consolidated against high shares of (one-sided) biological regulation. The statistical power of downstream analyses can be increased by focusing on Normics-selected subsets of high DE likelihood.

Droplet Microarray as a Powerful Platform for Seeking New Antibiotics Against Multidrug-Resistant Bacteria.

Multidrug-resistant (MDR) bacteria is a severe threat to public health. Therefore, it is urgent to establish effective screening systems for identifying novel antibacterial compounds. In this study, a highly miniaturized droplet microarray (DMA) based high-throughput screening system is established to screen over 2000 compounds for their antimicrobial properties against carbapenem-resistant Klebsiella pneumoniae and methicillin resistant Staphylococcus aureus (MRSA). The DMA consists of an array of hydrophilic spots divided by superhydrophobic borders. Due to the differences in the surface wettability between the spots and the borders, arrays of hundreds of nanoliter-sized droplets containing bacteria and different drugs can be generated for screening applications. A simple colorimetric viability readout utilizing a conventional photo scanner is developed for fast single-step detection of the inhibitory effect of the compounds on bacterial growth on the whole array. Six hit compounds, including coumarins and structurally simplified estrogen analogs are identified in the primary screening and validated with minimum inhibition concentration assay for their antibacterial effect. This study demonstrates that the DMA-based high-throughput screening system enables the identification of potential antibiotics from novel synthetic compound libraries, offering opportunities for development of new treatments against multidrug-resistant bacteria.

"Cells-to-cDNA on Chip": Phenotypic Assessment and Gene Expression Analysis from Live Cells in Nanoliter Volumes Using Droplet Microarrays.

In vitro cell-based experiments are particularly important in fundamental biological research. Microscopy-based readouts to identify cellular changes in response to various stimuli are a popular choice, but gene expression analysis is essential to delineate the underlying molecular dynamics in cells. However, cell-based experiments often suffer from interexperimental variation, especially while using different readout methods. Therefore, establishment of platforms that allow for cell screening, along with parallel investigations of morphological features, as well as gene expression levels, is crucial. The droplet microarray (DMA) platform enables cell screening in hundreds of nanoliter droplets. In this study, a "Cells-to-cDNA on Chip" method is developed enabling on-chip mRNA isolation from live cells and conversion to cDNA in individual droplets of 200 nL. This novel method works efficiently to obtain cDNA from different cell numbers, down to single cell per droplet. This is the first established miniaturized on-chip strategy that enables the entire course of cell screening, phenotypic microscopy-based assessments along with mRNA isolation and its conversion to cDNA for gene expression analysis by real-time PCR on an open DMA platform. The principle demonstrated in this study sets a beginning for myriad of possible applications to obtain detailed information about the molecular dynamics in cultured cells.

Methods for the frugal labeler: Multi-class semantic segmentation on heterogeneous labels.

Deep learning increasingly accelerates biomedical research, deploying neural networks for multiple tasks, such as image classification, object detection, and semantic segmentation. However, neural networks are commonly trained supervised on large-scale, labeled datasets. These prerequisites raise issues in biomedical image recognition, as datasets are generally small-scale, challenging to obtain, expensive to label, and frequently heterogeneously labeled. Furthermore, heterogeneous labels are a challenge for supervised methods. If not all classes are labeled for an individual sample, supervised deep learning approaches can only learn on a subset of the dataset with common labels for each individual sample; consequently, biomedical image recognition engineers need to be frugal concerning their label and ground truth requirements. This paper discusses the effects of frugal labeling and proposes to train neural networks for multi-class semantic segmentation on heterogeneously labeled data based on a novel objective function. The objective function combines a class asymmetric loss with the Dice loss. The approach is demonstrated for training on the sparse ground truth of a heterogeneous labeled dataset, training within a transfer learning setting, and the use-case of merging multiple heterogeneously labeled datasets. For this purpose, a biomedical small-scale, multi-class semantic segmentation dataset is utilized. The heartSeg dataset is based on the medaka fish's position as a cardiac model system. Automating image recognition and semantic segmentation enables high-throughput experiments and is essential for biomedical research. Our approach and analysis show competitive results in supervised training regimes and encourage frugal labeling within biomedical image recognition.

Simple assessment of viability in 2D and 3D cell microarrays using single step digital imaging.

Simple and rapid imaging and analysis of 2D and 3D cell culture compatible with miniaturized arrays of nanoliter droplets are essential for high-throughput screening and personalized medicine applications. In this study, we have developed a simple one-step, cost-effective and sensitive colorimetric method for the analysis of cell viability in 2D and 3D cell cultures on a nanoliter droplet microarray. The method utilizes a flatbed document scanner that detects a color change in response to cell metabolism in nanoliter droplets with high sensitivity in a single step without the need for expensive specialized equipment. This new nanoliter-based method is faster and more sensitive than equivalent methods using multi-well plate assays. The method detects quantifiable signal from as few as 10 cells and requires only 5 min. This is 2.5 to 10-fold more sensitive and 12 times faster than the same assay in multi-well plates. The method is simple, affordable, fast and sensitive. It can be used for various applications including high-throughput cell-based and biochemical screenings.

Equilibrium droplet shapes on chemically patterned surfaces: theoretical calculation, phase-field simulation, and experiments.

HYPOTHESIS: Droplet wetting on a solid substrate is affected by the surface heterogeneity. Introducing patterned wettability on the solid substrate is expected to engender anisotropic wetting morphologies, thereby manipulating droplet wetting behaviors. However, when the droplet size is comparable with that of the surface heterogeneity, the wetting morphologies cannot be depicted by the quintessential Cassie's theory but should be possible to be predicted from the perspective of thermodynamics via surface energy minimization. METHODS: Here, we investigate the equilibrium droplet shapes on chemically patterned substrates by using an analytical model, phase-field simulations, and experiments. The former two methods are sharp and diffuse interface treatments, respectively, which both are based on minimizing the free energy of the system. The experimental results are obtained by depositing droplets on chemically patterned glass substrates. FINDINGS: Various anisotropic wetting shapes are found from the three methods. Excellent agreement is observed between different methods, showing the possibility to quantify the anisotropic wetting droplet morphologies on patterned substrates by present methods. We also address a series of non-rotationally symmetric droplet shapes, which is the first resport about these special wetting morphologies. Furthermore, we reveal the anisotropic wetting shapes in a quasi-equilibrium evaporation process.

[Pilot study examining a profession-oriented rehabilitation concept for nursing professions]. / Pilotstudie zur Evaluation einer medizinisch-beruflich orientierten Rehabilitation für Pflegekräfte.

Pilot study examining a profession-oriented rehabilitation concept for nursing professions Objectives: Nursing professions are associated with high levels of psychological distress, high numbers of absent days and premature retirement. To achieve higher return-to-work rates, psychosomatic rehabilitation is expected to offer treatments tailored to workplace demands. This pilot study is the first to examine the effects of a new workplace-oriented medical rehabilitation program for nursing professions. Methods: A total of N = 145 depressed patients in nursing occupations (86.9 % female; 50.9 ± 7.34 years) took part in a workplace-oriented rehabilitation program for nursing professions. At admission they were compared to N = 147 depressed patients (63.27 % female; 49.36 ± 7.58 years) in non-nursing professions regarding patterns of work-related experience and behaviour (AVEM) using a MANOVA with follow-up ANOVAs for individual subscales. Changes in work-related attitudes among the nursing professions following completion of the intervention were assessed using a MANOVA followed by repeated measures ANOVAs. The effect of the workplace- oriented intervention on depressiveness (BDI-II) was compared to a treatment program for depression using a mixed model after taking potentially confounding variables into account. Results: At entry, depressed patients in nursing professions scored significantly higher on AVEM scale willingness to work to exhaustion and lower on AVEM scale distancing ability compared to depressed patients in other professions. Following completion of the workplace-oriented intervention program for nursing professions, participants showed a significant reduction on AVEM scales subjective importance of work, willingness to work to exhaustion, and striving for perfection. Increasing scores were observed on the distancing ability and life satisfaction scales. Depression scores had significantly improved at discharge in both participants of the work-oriented intervention and the disorder-specific intervention for depressive disorders, whereas neither group differences nor interaction effects were found. Conclusions: The work-oriented intervention for nursing professions successfully induced changes in maladaptive work-related attitudes. Improvements in depressiveness did not significantly differ from an intervention targeting depression specifically.

Machine learning methods for automated classification of tumors with papillary thyroid carcinoma-like nuclei: A quantitative analysis.

When approaching thyroid gland tumor classification, the differentiation between samples with and without "papillary thyroid carcinoma-like" nuclei is a daunting task with high inter-observer variability among pathologists. Thus, there is increasing interest in the use of machine learning approaches to provide pathologists real-time decision support. In this paper, we optimize and quantitatively compare two automated machine learning methods for thyroid gland tumor classification on two datasets to assist pathologists in decision-making regarding these methods and their parameters. The first method is a feature-based classification originating from common image processing and consists of cell nucleus segmentation, feature extraction, and subsequent thyroid gland tumor classification utilizing different classifiers. The second method is a deep learning-based classification which directly classifies the input images with a convolutional neural network without the need for cell nucleus segmentation. On the Tharun and Thompson dataset, the feature-based classification achieves an accuracy of 89.7% (Cohen's Kappa 0.79), compared to the deep learning-based classification of 89.1% (Cohen's Kappa 0.78). On the Nikiforov dataset, the feature-based classification achieves an accuracy of 83.5% (Cohen's Kappa 0.46) compared to the deep learning-based classification 77.4% (Cohen's Kappa 0.35). Thus, both automated thyroid tumor classification methods can reach the classification level of an expert pathologist. To our knowledge, this is the first study comparing feature-based and deep learning-based classification regarding their ability to classify samples with and without papillary thyroid carcinoma-like nuclei on two large-scale datasets.

Long photoperiod impairs learning in male but not female medaka.

Day length in conjunction with seasonal cycles affects many aspects of animal biology. We have studied photoperiod-dependent alterations of complex behavior in the teleost, medaka (Oryzias latipes), a photoperiodic breeder, in a learning paradigm whereby fish have to activate a sensor to obtain a food reward. Medaka were tested under a long (14:10 LD) and short (10:14 LD) photoperiod in three different groups: mixed-sex, all-males, and all-females. Under long photoperiod, medaka mixed-sex groups learned rapidly with a stable response. Unexpectedly, males-only groups showed a strong learning deficit, whereas females-only groups performed efficiently. In mixed-sex groups, female individuals drove group learning, whereas males apparently prioritized mating over feeding behavior resulting in strongly reduced learning performance. Under short photoperiod, where medaka do not mate, male performance improved to a level similar to that of females. Thus, photoperiod has sex-specific effects on the learning performance of a seasonal vertebrate.