Bioimage analysis of fluorescent labels is widely used in the life sciences. Recent advances in deep learning (DL) allow automating time-consuming manual image analysis processes based on annotated training data. However, manual annotation of fluorescent features with a low signal-to-noise ratio is somewhat subjective. Training DL models on subjective annotations may be instable or yield biased models. In turn, these models may be unable to reliably detect biological effects. An analysis pipeline integrating data annotation, ground truth estimation, and model training can mitigate this risk. To evaluate this integrated process, we compared different DL-based analysis approaches. With data from two model organisms (mice, zebrafish) and five laboratories, we show that ground truth estimation from multiple human annotators helps to establish objectivity in fluorescent feature annotations. Furthermore, ensembles of multiple models trained on the estimated ground truth establish reliability and validity. Our research provides guidelines for reproducible DL-based bioimage analyses.

Survival in a natural environment forces an individual into constantly adapting purposive behavior. Specified interoceptive neurons monitor metabolic and physiological balance and activate dedicated brain circuits to satisfy essential needs, such as hunger, thirst, thermoregulation, fear, or anxiety. Neuropeptides are multifaceted, central components within such life-sustaining programs. For instance, nutritional depletion results in a drop in glucose levels, release of hormones, and activation of hypothalamic and brainstem neurons. These neurons, in turn, release several neuropeptides that increase food-seeking behavior and promote food intake. Similarly, internal and external threats activate neuronal pathways of avoidance and defensive behavior. Interestingly, specific nuclei of the hypothalamus and extended amygdala are activated by both hunger and fear. Here, we introduce the relevant neuropeptides and describe their function in feeding and emotional-affective behaviors. We further highlight specific pathways and microcircuits, where neuropeptides may interact to identify prevailing homeostatic needs and direct respective compensatory behaviors. A specific focus will be on neuropeptide Y, since it is known for its pivotal role in metabolic and emotional pathways. We hypothesize that the orexigenic and anorexigenic properties of specific neuropeptides are related to their ability to inhibit fear and anxiety.

Disproportionate, maladapted, and generalized fear are essential hallmarks of posttraumatic stress disorder (PTSD), which develops upon severe trauma in a subset of exposed individuals. Among the brain areas that are processing fear memories, the hippocampal formation exerts a central role linking emotional-affective with cognitive aspects. In the hippocampus, neuronal excitability is constrained by multiple GABAergic interneurons with highly specialized functions and an extensive repertoire of co-released neuromodulators. Neuropeptide Y (NPY) is one of these co-transmitters that significantly affects hippocampal signaling, with ample evidence supporting its fundamental role in emotional, cognitive, and metabolic circuitries. Here we investigated the role of NPY in relation to GABA, both released from the same interneurons of the dorsal dentate gyrus (DG), in different aspects of fear conditioning. We demonstrated that activation of dentate GABA neurons specifically during fear recall reduced cue-related as well as trace-related freezing behavior, whereas inhibition of the same neurons had no significant effects. Interestingly, concomitant overexpression of NPY in these neurons did not further modify fear recall, neither under baseline conditions nor upon chemogenetic stimulation. However, potentially increased co-release of NPY substantially reduced contextual fear, promoted extinction learning, and long-term suppression of fear in a foreground context–conditioning paradigm. Importantly, NPY in the dorsal DG was not only expressed in somatostatin neurons, but also in parvalbumin-positive basket cells and axoaxonic cells, indicating intense feedback and feedforward modulation of hippocampal signaling and precise curtailing of neuronal engrams. Thus, these findings suggest that co-release of NPY from specific interneuron populations of the dorsal DG modifies dedicated aspects of hippocampal processing by sharpening the activation of neural engrams and the consecutive fear response. Since inappropriate and generalized fear is the major impediment in the treatment of PTSD patients, the dentate NPY system may be a suitable access point to ameliorate PTSD symptoms and improve the inherent disease course.

AbstractFluorescent labeling of biomolecules is widely used for bioimage analyses throughout the life sciences. Recent advances in deep learning (DL) have opened new possibilities to scale the image analysis processes through automation. However, the annotation of fluorescent features with a low signal-to-noise ratio is frequently based on subjective criteria. Training on subjective annotations may ultimately lead to biased DL models yielding irreproducible results. An end-to-end analysis process that integrates data annotation, ground truth estimation, and model training can mitigate this risk. To highlight the importance of this integrated process, we compare different DL-based analysis approaches. Based on data from different laboratories, we show that ground truth estimation from multiple human annotators is indispensable to establish objectivity in fluorescent feature annotations. We demonstrate that ensembles of multiple models trained on the estimated ground truth establish reliability and validity. Our research provides guidelines for reproducible and transparent bioimage analyses using DL methods.