@phdthesis{SchukraftgebScheffler2024,
  author    = {Schukraft [geb. Scheffler], Nina},
  title     = {Integrated defensive states and their neuronal correlates in the Periaqueductal Gray},
  doi       = {10.25972/OPUS-34745},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-347458},
  school      = {Universit{\"a}t W{\"u}rzburg},
  year      = {2024},
  abstract  = {In the face of threat, animals react with a defensive reaction to avoid or reduce harm. This defensive reaction encompasses apart from behavioral changes also physiological, analgetic, and endocrine adaptations. Nonetheless, most animal studies on fear and anxiety are based on behavioral observations only, disregarding other aspects of the defensive reaction, or integrating their inter-related dynamics only insufficiently. The first part of this thesis aimed in characterizing patterned associations of behavioral and physiological responses, termed integrated defensive states. Analyzing cardiac and behavioral responses in mice undergoing multiple fear and anxiety paradigms revealed a complex and dynamic interaction of those readouts on both, short and long timescales. Microstates, stereotypical combinations of i.e. freezing and decelerating heart rates, are short-lasting and were, in turn, shown to be influenced by slow acting macrostate changes. One of those higher order macrostates, called `rigidity`, was defined as a latent process that constrains the range of momentary displayed heart rate values. Furthermore, integrated defensive states were found to be highly dependent on the cue and the context the animals are confronted with. Importantly, same behavioral observations, i.e. freezing, were associated with distinct cardiac responses, highlighting the importance of multivariate analysis of integrated defensive states. Defensive states are orchestrated by the brain, which has evolved evolutionary conserved survival circuits. A central brain area of these circuits is the periaqueductal gray (PAG) in the midbrain. It plays a pivotal role in mediating defensive states, as it receives signals about external and internal information from multiple brain regions and sends information to both, higher order brain areas as well as to the brainstem ultimately causing the execution of threat responses. In the second part of this thesis, different neuronal circuit elements in the PAG were optically manipulated in order to gain mechanistic insight into the defense network in the brain underlying the previously delineated cardio-behavioral defensive states. Optical activation of glutamatergic PAG neurons evoked heterogeneous, light-intensity dependent responses. However, a further molecular restriction of the glutamatergic neuronal population targeting only Chx10+ neurons, led to a cardio-behavioral state that resembled spontaneous freezing-bradycardia bouts. In summary, this thesis presents a multivariate description of defensive states, which includes the complex interaction of cardiac and behavioral responses on different timescales and, furthermore, functionally dissects different excitatory and inhibitory PAG circuit elements mediating these defensive states.},
  subject      = {Perianova, Irina},
  language  = {en}
}
@article{GriebelSegebarthSteinetal.2023,
  author    = {Griebel, Matthias and Segebarth, Dennis and Stein, Nikolai and Schukraft, Nina and Tovote, Philip and Blum, Robert and Flath, Christoph M.},
  title     = {Deep learning-enabled segmentation of ambiguous bioimages with deepflash2},
  series = {Nature Communications},
  volume    = {14},
  journal   = {Nature Communications},
  doi       = {10.1038/s41467-023-36960-9},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-357286},
  year      = {2023},
  abstract  = {Bioimages frequently exhibit low signal-to-noise ratios due to experimental conditions, specimen characteristics, and imaging trade-offs. Reliable segmentation of such ambiguous images is difficult and laborious. Here we introduce deepflash2, a deep learning-enabled segmentation tool for bioimage analysis. The tool addresses typical challenges that may arise during the training, evaluation, and application of deep learning models on ambiguous data. The tool's training and evaluation pipeline uses multiple expert annotations and deep model ensembles to achieve accurate results. The application pipeline supports various use-cases for expert annotations and includes a quality assurance mechanism in the form of uncertainty measures. Benchmarked against other tools, deepflash2 offers both high predictive accuracy and efficient computational resource usage. The tool is built upon established deep learning libraries and enables sharing of trained model ensembles with the research community. deepflash2 aims to simplify the integration of deep learning into bioimage analysis projects while improving accuracy and reliability.},
  language  = {en}
}