ICoRD: Iterative Correlation-Based ROI Detection Method for the Extraction of Neural Signals in Calcium Imaging

In vivo calcium imaging is a standard neuroimaging technique that allows the simultaneous observation of neuronal population activity. In calcium imaging, the activation signals of neurons are key information for the investigation of neural circuits. For efficient extraction of the calcium signals of neurons, selective detection of the region of interest (ROI) pixels corresponding to the active subcellular region of the target neuron is essential. However, current ROI detection methods for calcium imaging data exhibit relatively low extraction performance from neurons with a low signal-to-noise power ratio (SNR). This is problematic because a low SNR is unavoidable in many biological experimental settings. Therefore, we propose an iterative correlation-based ROI detection (ICoRD) method that robustly extracts the calcium signal of the target neuron from a calcium imaging series with severe noise. ICoRD extracts calcium signals closer to the ground truth than the conventional method from simulated calcium imaging data in all low SNR ranges. Additionally, this study confirmed that ICoRD robustly extracts activation signals against noise, even within in vivo environments. ICoRD showed reliable detection from neurons with low SNR and sparse activation, which were not detected by the conventional methods. ICoRD will facilitate our understanding of neural circuit activity by providing significantly improved ROI detection from noisy images.

Imaging microvasculature network evolution and neurodegeneration with precise photothrombosis approach

In the cerebral cortex, the vasculature plays important homeostatic functions, especially at the highly connected complex capillary networks. The association of focal capillary ischemia with the neurodegenerative disease as well as the laminar vascular dynamics have prompted studies of vascular micro-occlusion via photothrombosis. However, technical challenges of this approach remain, including increased temporal precision of occlusion, increasing the depth of vascular occlusion, understanding how such micro-occlusion impacts local blood flow, and ultimately the neuronal effects of such changes. Here, we have developed a novel approach that employs ultra-fast multiphoton light to induce focal Rose Bengal-induced photothrombosis. We demonstrated induction of highly precise and fast occlusion of microvessels at various types and depths. The change of the microvascular architecture and hemodynamics after occlusion revealed the autoregulation and significant difference between upstream vs downstream in layer 2/3. Further, we found that micro-occlusion at two different layers within the same vascular arbor results in distinct effects on the acute flow redistribution mechanism. To examine neuronal effects of such micro-occlusion, we produced infarct of capillaries surrounding a labeled target neuron and found this induces dramatic and rapid lamina-specific degeneration in neuronal dendritic architecture. In sum, our technique enhanced the precision and power of the photothrombotic study of microvascular function. The current results pointed to the importance of laminar scale regulation within the microvascular network, a finding which may be relevant for models of neurovascular disease.

PatchPerPixMatch for Automated 3d Search of Neuronal Morphologies in Light Microscopy

Studies of individual neurons in the Drosophila nervous system are facilitated by transgenic lines that sparsely and repeatably label respective neurons of interest. Sparsity can be enhanced by means of intersectional approaches like the split-GAL4 system, which labels the positive intersection of the expression patterns of two (denser) GAL4 lines. To this end, two GAL4 lines have to be identified as labelling a neuron of interest. Current approaches to tackling this task include visual inspection, as well as automated search in 2d projection images, of single cell multi-color flip-out (MCFO) acquisitions of GAL4 expression patterns. There is to date no automated method available that performs full 3d search in MCFO imagery of GAL4 lines, nor one that leverages automated reconstructions of the labelled neuron morphologies. To close this gap, we propose PatchPerPixMatch, a fully automated approach for finding a given neuron morphology in MCFO acquisitions of Gen1 GAL4 lines. PatchPerPixMatch performs automated instance segmentation of MCFO acquisitions, and subsequently searches for a target neuron morphology by minimizing an objective that aims at covering the target with a set of well-fitting segmentation fragments. PatchPerPixMatch is computationally efficient albeit being full 3d, while also highly robust to inaccuracies in the automated neuron instance segmentation. We are releasing PatchPerPixMatch search results for ~30,000 neuron morphologies from the Drosophila hemibrain in ~20,000 MCFO acquisitions of ~3,500 Gen1 GAL4 lines.

Stepwise neuronal network pattern formation in agarose gel during cultivation using non-destructive microneedle photothermal microfabrication

Conventional neuronal network pattern formation techniques cannot control the arrangement of axons and dendrites because network structures must be fixed before neurite differentiation. To overcome this limitation, we developed a non-destructive stepwise microfabrication technique that can be used to alter microchannels within agarose to guide neurites during elongation. Micropatterns were formed in thin agarose layer coating of a cultivation dish using the tip of a 0.7 $$\upmu \mathrm{m}$$ μ m -diameter platinum-coated glass microneedle heated by a focused 1064-nm wavelength infrared laser, which has no absorbance of water. As the size of the heat source was 0.7 $$\upmu \mathrm{m}$$ μ m , which is smaller than the laser wavelength, the temperature fell to 45 $$^\circ \hbox {C}$$ ∘ C within a distance of 7.0 $$\upmu \mathrm{m}$$ μ m from the edge of the etched agarose microchannel. We exploited the fast temperature decay property to guide cell-to-cell connection during neuronal network cultivation. The first neurite of a hippocampal cell from a microchamber was guided to a microchannel leading to the target neuron with stepwise etching of the micrometer resolution microchannel in the agarose layer, and the elongated neurites were not damaged by the heat of etching. The results indicate the potential of this new technique for fully direction-controlled on-chip neuronal network studies.

Probabilistic Spike Propagation for Efficient Hardware Implementation of Spiking Neural Networks

Spiking neural networks (SNNs) have gained considerable attention in recent years due to their ability to model temporal event streams, be trained using unsupervised learning rules, and be realized on low-power event-driven hardware. Notwithstanding the intrinsic desirable attributes of SNNs, there is a need to further optimize their computational efficiency to enable their deployment in highly resource-constrained systems. The complexity of evaluating an SNN is strongly correlated to the spiking activity in the network, and can be measured in terms of a fundamental unit of computation, viz. spike propagation along a synapse from a single source neuron to a single target neuron. We propose probabilistic spike propagation, an approach to optimize rate-coded SNNs by interpreting synaptic weights as probabilities, and utilizing these probabilities to regulate spike propagation. The approach results in 2.4–3.69× reduction in spikes propagated, leading to reduced time and energy consumption. We propose Probabilistic Spiking Neural Network Application Processor (P-SNNAP), a specialized SNN accelerator with support for probabilistic spike propagation. Our evaluations across a suite of benchmark SNNs demonstrate that probabilistic spike propagation results in 1.39–2× energy reduction with simultaneous speedups of 1.16–1.62× compared to the traditional model of SNN evaluation.

Short-term dynamics of long-range corticocortical synapses revealed by selective optical stimulation

Synapses are continually regulated by their own activity. In the neocortex, direct interactions between cortical areas play a central role in cognitive function, but the dynamic regulation of these long-range corticocortical synapses by activity and their impact on a postsynaptic target neuron is unclear. Here, we use an optogenetic strategy to study the connections between mouse somatosensory and motor cortex. We found that short-term synaptic facilitation was strong in both corticocortical synapses, resulting in far more sustained responses than local intra-cortical and thalamocortical connections. This facilitation was dependent on the presynaptic calcium sensor synaptotagmin-7 and altered by several optogenetic approaches. Recordings revealed that during repetitive activation, the short-term dynamics of corticocortical synapses enhanced the excitability of layer 2/3 pyramidal neurons, increasing the probability of spiking with activity. Furthermore, the properties of the connections linking primary with secondary somatosensory cortex resemble those between somatosensory-motor areas. These results reveal a synaptic mechanism by which corticocortical projections may mediate specific changes in cellular excitability over relatively extended periods.

HGG-02. NEUROPHYSIOLOGICAL SMALL MOLECULE SCREEN TO TARGET NEURON-GLIOMA INTERACTIONS IN PEDIATRIC HIGH-GRADE GLIOMAS

Neurons stimulate glioma growth via synaptic and paracrine signaling mechanisms. We recently demonstrated that neurons form AMPA receptor-dependent synapses with glioma cells, and that neuronal activity also induces potassium-evoked currents that are amplified by gap junctions coupling glioma cells. However, our understanding of the neurotransmitters, receptors, and ion channels participating in neuron-glioma signaling remains incomplete. We have recently developed a high-throughput neuron-glioma co-culture strategy to screen small molecules for agents that may disrupt neuron-glioma signaling. Glioma cell proliferation is increased tenfold when cultured together with neurons; this robust biological effect can be probed in a targeted screen of compounds influencing neurotransmitter receptors and ion channels. The neurophysiological small molecule library used was curated to include approved anti-epileptics, neuroleptics, and antidepressants, as well as a variety of other compounds acting on different neurotransmitter types and ion channels. Hits from the primary screen were run through a counter-screen using glioma cells grown alone without neurons, to identify compounds that specifically affect neuron-glioma interactions. Correlation of the screening results with drug mechanisms of action will allow us to map out the key neurotransmitter pathways regulating glioma growth, which can be further validated using genetic and in vivo experiments. Drugs identified in this glioma neuroscience screen may be readily translated into much-needed therapeutics for children with high-grade glioma.

Robust switches in thalamic network activity require a timescale separation between sodium and T-type calcium channel activations

Switches in brain states, synaptic plasticity and neuromodulation are fundamental processes in our brain that take place concomitantly across several spatial and timescales. All these processes target neuron intrinsic properties and connectivity to achieve specific physiological goals, raising the question of how they can operate without interfering with each other. Here, we highlight the central importance of a timescale separation in the activation of sodium and T-type calcium channels to sustain robust switches in brain states in thalamic neurons that are compatible with synaptic plasticity and neuromodulation. We quantify the role of this timescale separation by comparing the robustness of rhythms of six published conductance-based models at the cellular, circuit and network levels. We show that robust rhythm generation requires a T-type calcium channel activation whose kinetics are situated between sodium channel activation and T-type calcium channel inactivation in all models despite their quantitative differences.

Spike neuron optimization using deep reinforcement learning

Deep reinforcement learning (DRL) which involved reinforcement learning and artificial neural network allows agents to take the best possible actions to achieve goals. Spiking Neural Network (SNN) faced difficulty in training due to the non-differentiable spike function of spike neuron. In order to overcome the difficulty, Deep Q network (DQN) and Deep Q learning with normalized advantage function (NAF) are proposed to interact with a custom environment. DQN is applied for discrete action space whereas NAF is implemented for continuous action space. The model is trained and tested to validate its performance in order to balance the firing rate of excitatory and inhibitory population of spike neuron by using both algorithms. Training results showed both agents able to explore in the custom environment with OpenAI Gym framework. The trained model for both algorithms capable to balance the firing rate of excitatory and inhibitory of the spike neuron. NAF achieved 0.80% of the average percentage error of rate of difference between target and actual neuron rate whereas DQN obtained 0.96%. NAF attained the goal faster than DQN with only 3 steps taken for actual output neuron rate to meet with or close to target neuron firing rate.