Journal of Neuroscience Methods

Currently, implantation of electroencephalogram (EEG) electrodes in laboratory animals is time-consuming and requires specialized equipment. We present a novel method for EEG recordings in mice that utilizes thin needle electrodes. These electrodes are inserted into the skull at predetermined locations by gently pressing them against the bone surface. To ensure stable fixation of the implant, hook-shaped needles are positioned along the lateral aspects of the skull. The electrodes are connected to a multipin connector and secured to the skull using dental composite, after which the animal is allowed to recover from anesthesia. Importantly, procedures such as skull drilling and screw placement are not required, allowing the entire surgery to be completed in less than 15 minutes. Consequently, this EEG implantation approach is rapid and minimally invasive. Results of our studies indicate that EEG recordings obtained with needle electrodes are not inferior to those obtained with screw electrodes. Overall, the method is designed to enhance the accuracy and efficiency of EEG recording studies while improving animal welfare. O_LISimplifies the placement of EEG electrodes. C_LIO_LIReduces the time required for electrode implantation. C_LI Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=67 SRC="FIGDIR/small/715731v1_ufig1.gif" ALT="Figure 1"> View larger version (44K): org.highwire.dtl.DTLVardef@e5608org.highwire.dtl.DTLVardef@1325ea4org.highwire.dtl.DTLVardef@1e37202org.highwire.dtl.DTLVardef@1521bb8_HPS_FORMAT_FIGEXP M_FIG C_FIG

Electroencephalography (EEG) data are inherently contaminated by non-neuronal noise, including eye movements, muscle activity, cardiac signals, electrical interference, and technical issues such as poorly connected electrodes. Preprocessing to remove these artefacts is essential, yet the optimal method remains unclear due to the vast number of available techniques, their combinatorial use in pipelines, and adjustable parameters. Consequently, most studies adopt ad hoc preprocessing strategies based on dataset characteristics, study goals, and researcher expertise, with little justification for their choices. Such variability can influence downstream results, potentially determining whether effects are detected, and introduces risks of questionable analytical practices. Here, we present a method to objectively evaluate and compare preprocessing pipelines. Our approach uses realistically simulated signals injected into real EEG data as "ground truth", enabling the assessment of a pipelines ability to remove noise without distorting neuronal signals. This evaluation is independent of the studys main analyses, ensuring that pipeline selection does not bias results. By applying this procedure, researchers can select preprocessing strategies that maximize signal-to-noise ratio while maintaining the integrity of the neural signal, improving both reproducibility and interpretability of EEG studies. Although the data presented here focuses on processing and analysis most relevant for ERP research, the method can be flexibly expanded to other types of analyses or signals.

Chronic low back pain (CLBP) is persistent and refractory, affecting 20-30% of population worldwide. Neurofeedback has been explored as a potential non-pharmacological intervention for chronic pain, although evidence in CLBP remains limited. This study evaluated PainWaive, a consumer-grade digitally-delivered neurofeedback intervention targeting multiple pain-related frequency bands recorded over the sensorimotor cortex in individuals with CLBP. In a multiple-baseline experimental design, four participants completed daily assessments of pain severity and pain interference during randomly-assigned baseline phases of 7, 10, 14, and 20 days, followed by 20 sessions of the PainWaive intervention over four weeks. Daily pain assessments continued during the post-intervention and follow-up phases. Participants rated PainWaive's usability and acceptability at post-intervention. Anxiety, depression, wellbeing, and sleep disturbance were assessed at three timepoints. Aggregated Tau-U analyses indicated a large effect (-0.67) on pain severity from baseline to intervention and very large from baseline to post-intervention (-0.92) and follow-up (-0.92) phases. Large effects (-0.63, -0.62, and -0.70) were also observed for pain interference. Individual-level analyses showed significant reductions across all participants, with visual inspection confirming progressive decreases over time. The intervention was rated usable and acceptable by all participants, while psychological outcomes were mixed and varied across participants. The findings provide promising evidence that the PainWaive neurofeedback intervention may reduce pain severity and pain interference in some individuals with CLBP. By prioritising accessibility, usability, and self-administration, PainWaive supports a foundation for more patient-centred, technology-enabled approaches to chronic pain management. Further evaluation of this approach in randomised trials is required to establish efficacy.

BackgroundPrimary astrocyte cultures derived from neonatal rodent cortices provide a controlled system for investigating astrocyte-specific mechanisms. However, mixed glial preparations frequently contain contaminating microglia and oligodendrocyte progenitor cells, and most existing protocols require pooling tissue from multiple mouse pups to obtain sufficient astrocyte yields. This approach is impractical as it obscures sex and genotype, limits investigations of sex dependent astrocyte phenotypes, and precludes studies in certain transgenic models. To address this gap, our protocol achieves a high astrocyte yield from a single neonatal mouse brain, enabling sex- and genotype-specific cultures without the need for pooling. Mechanical removal of oligodendrocyte progenitors combined with pharmacological depletion of microglia using a Colony Stimulating Factor 1 Receptor (CSF1R) inhibitor produces highly enriched astrocytes suitable for functional assays, including those focused on sex-specific biology. MethodsCortical tissue was isolated from a single mouse pup is mechanically dissociated in astrocyte media. Cell suspensions are transferred to poly-D-lysine-coated flasks in astrocyte media. After 10-15 days in culture, OPCs are mechanically removed by horizontal shaking and microglia are selectively depleted by incubating cultures with CSF1R inhibitor PLX5622 for 24, 48, 72 and 96 hours. After PLX treatment, media is replaced and enriched astrocytes were maintained or passaged for experimentation. The sex of the pups is determined by PCR performed on DNA extracted from tail biopsies. ResultsImmunocytochemical analysis for astrocyte and microglia markers (GFAP and Iba1, respectively) showed that 24 hours of PLX5622 treatment did not fully eliminate microglia from mixed glial cultures. Extending treatment to 48 hours effectively depleted microglia while minimizing cytotoxicity and astrocyte loss and produced a pure, high-yield, sex-specific primary astrocyte culture. PCR reliably enabled the sex identification of pups used in culture using DNA extracted from tail biopsies. DiscussionThis protocol provides an efficient and reproducible method for generating high-purity, sex-specific primary astrocyte cultures from a single mouse brain. It improves consistency and purity while eliminating the need to pool tissue, preserving sex and genotype and enabling studies in transgenic mouse lines of both sexes.

The use of domestic pigs in clinical training and biomedical research is expanding rapidly, increasing the need for reliable, noninvasive indicators of health and welfare. Vocal analysis offers a non-invasive promising tool, yet the acoustic repertoire of adult domestic pigs remains poorly defined. However, the vocalization repertoire of adult domestic pigs has yet to be characterized. This study characterizes the vocal repertoire of adult pigs housed in a biomedical research laboratory. Twelve mixed-breed pigs (2-3 months old; 5 males, 7 females) were recorded during routine husbandry and experimental procedures. Vocal classification was conducted using perceptual and objective clustering techniques. First, aural- visual (AV) inspection of spectrograms was used to construct a hierarchical repertoire. Second, a two-step cluster analysis based on six acoustic parameters (5% frequency, first quartile frequency, center frequency, 90% bandwidth, interquartile range bandwidth, and 90% duration) provided an objective classification. Agreement between methods was evaluated using Cramers V. A total of 1,136 vocalizations from 69 recordings were analyzed. AV classification revealed five major vocal classes-- grunt, squeal, complex, scream, and bark--subdividing into 16 distinct call types. Standardized definitions integrating descriptive and quantitative criteria are provided. The two-step cluster analysis identified two clusters as the optimal statistical solution, with moderate agreement between methods (Cramers V = 0.67, p < 0.0001). Most AV-defined call types aligned with previously reported repertoires, although whines, yelps, and stable screams were unique to this study. While two-cluster solutions are commonly reported, our findings indicate that richer acoustic structure exists and that high gradation among pig calls may limit the resolution of statistical clustering. These results establish a detailed acoustic framework for adult pig vocalizations and provide essential groundwork for developing predictive models to enhance welfare assessment and support comparative research in laboratory-housed pigs.

Motion artifacts remain a barrier to in vivo calcium imaging in Drosophila melanogaster larvae. Here, we evaluate a multimodal immobilization approach that combines a Pluronic F-127 (PF-127) hydrogel with brief diethyl ether vapor exposure (5 minutes, 25{degrees}C) and compare it against hydrogel-only immobilization using custom MATLAB-based analysis software that performs NoRMCorre rigid motion correction. In wide-field GFP recordings at 1 Hz over approximately 60 minutes (N = 15 per group), the multimodal condition significantly reduced motion across all three core metrics after FDR correction (all q < 0.001), with large effect sizes for mean speed (Hedges g = -1.18) and median step size (g = -1.36). In a secondary analysis of the first 30 minutes, uniformly large effect sizes (|g| = 1.10-1.51) were observed, consistent with stronger initial chemical immobilization that partially wanes over the recording period. We implemented a dual-flag quality control system that distinguishes motion data reliability from ROI detection eligibility. Control calcium recordings (33.33 Hz, [~]5 minutes; N = 23) yielded 368 ROIs with a mean SNR 30.4 {+/-} 16.9 and an event rate of 0.228 {+/-} 0.113 Hz. Experimental recordings (N = 21) yielded 295 ROIs with SNR 18.0 {+/-} 10.6 and event rate 0.309 {+/-} 0.188 Hz. SNR was higher in controls (Cliffs{delta} = 0.50, p < 0.001), while event rate was modestly higher in the experimental group at the ROI level ({delta} = -0.22, p < 0.001), though this difference did not reach significance at the sample level, suggesting altered but not suppressed calcium dynamics. These results support a practical, accessible immobilization workflow for larval calcium imaging. HighlightsO_LIBrief ether + hydrogel approach reduces larval motion 85-91% vs. hydrogel alone C_LIO_LIDual-flag QC system separates motion reliability from calcium ROI eligibility C_LIO_LICalcium event rates not suppressed under multimodal immobilization C_LIO_LIComplete MATLAB pipeline for motion analysis and calcium imaging provided C_LIO_LIAccessible protocol requires only standard laboratory supplies C_LI

Electroencephalography and magnetoencephalography (EEG/MEG) provide non-invasive measurements of large-scale neural activity but do not directly reveal the underlying cortical sources, motivating the use of source localisation algorithms. However, objective evaluation of these methods remains challenging due to the absence of an experimentally verifiable ground truth. This study presents a simulation framework for generating biologically plausible ictal dynamics and their corresponding EEG signals to enable systematic benchmarking of source imaging approaches. Cortical seizure initiation and propagation were simulated using network-coupled neural mass (Epileptor) models, and combined with realistic forward models of the human head to produce macroscopic, electrophysiological data with known ground truth under varying conditions. Using this dataset, we evaluated established source localisation methods across idealised and realistic scenarios. Existing approaches achieved reasonable spatial accuracy under high-density, noise-free conditions; however, performance degraded substantially with reduced sensor coverage and added noise. This degradation was driven primarily by failures to recover source polarity, even when spatial localisation remained relatively accurate. These results suggest that current methods may be sufficient for identifying epileptogenic regions or tracking regional recruitment, but highlight polarity reconstruction as a key limitation for studies of seizure dynamics and network organisation. The proposed framework provides a reproducible and biologically grounded testbed for the development and evaluation of electrophysiological source localisation techniques.

Functional near infrared spectroscopy (fNIRS) is increasingly used in hearing and communication research, with advantages such as robustness to movement artifacts, improved spatial resolution, and flexibility of contexts in which it can be applied. At the same time, the field is progressively moving towards more continuous, naturalistic listening paradigms resulting in the widespread adoption of speech tracking analyses such as temporal response functions (TRFs) in electroencephalography (EEG) and magnetoencephalography (MEG) studies. However, it remains unclear whether these analyses can be applied to slower haemodynamic signals measured by fNIRS. In the present study, we investigated whether a TRF framework can similarly be applied to fNIRS data recorded during continuous speech perception. Eight participants listened to speech simultaneously while fNIRS signals were acquired in a hyperscanning setup. Speech features were regressed onto the haemodynamic responses to test the feasibility and interpretability of fNIRS-based TRFs. Prediction correlations between observed and modelled fNIRS signals across speech features were higher than those typically reported for EEG- and comparable to those reported for MEG-TRF studies. Moreover, these correlations did not overlap with a null distribution generated from triallJmismatched fNIRS data, confirming statistical significance and were slightly greater than those obtained from a conventional GLM approach. Our findings support that TRF estimation method can yield meaningful and statistically significant responses from fNIRS data. HighlightsO_LITRF modelling can be meaningfully applied to fNIRS data acquired during speech listening tasks. C_LIO_LIPrediction correlations between actual and modelled fNIRS signals were above chance level, with values comparable to previous EEG/MEG studies. C_LIO_LITRFs explained more fNIRS variance than a conventional GLM approach. C_LI

BackgroundTranscranial Magnetic Stimulation (TMS) studies identify the Resting Motor Threshold (RMT) to calibrate stimulation intensity. However, this procedure is time-consuming and subject to variability. We developed an automated procedure to improve the efficiency and standardization of RMT determination. New methodWe developed an algorithm that measures MEP amplitudes and automatically adjusts stimulation intensity to determine the RMT. Experiment 1 compared this automated method with the manual procedure in terms of reliability and equivalence. Experiment 2 developed a "Fast" automated process, assessing it against both the manual and initial automated procedures. ResultsAcross both experiments the automated approach demonstrated excellent test-retest reliability and strong agreement with the manual method (Intraclass Correlation Coefficients [&ge;]0.95), giving estimates of RMT statistically equivalent to those of manual measurements within {+/-}3% MSO, with the majority of comparisons within {+/-}2% MSO. Experiment 2 optimized the procedure, allowing empirical determination of the RMT in an average of <3 minutes with only 33-34 pulses. Comparison with existing methods RMT-Finder provides a reliable and time-efficient alternative to manual approaches. To the best of our knowledge RMT-Finder presents the first closed-loop feedback approach to identify the RMT without manual intervention. This procedure can improve standardization and reproducibility in TMS studies. ConclusionsAutomating RMT assessment allows rapid and highly reproducible assessment of this standard TMS measurement, making it viable for inclusion in routine clinical applications that require standardized procedures.

1Markerless pose estimation has emerged as a powerful technique for animal behavior quantification, capable of high resolution tracking of body parts. Many neuroscience labs rely on tools like DeepLabCut and SLEAP, which provide accessible interfaces but restrict users to a narrow set of models and configurations. In this work, we adopt MMPose an open source, general-purpose computer vision library to build a workflow for training and evaluating multiple state-of-the-art models on animal video datasets. We benchmark these models in two scenarios: (1) a complex maze assay with occlusions and varied backgrounds, and (2) a simpler open field arena with a high-contrast background. Our results show that a bottomup model (DEKR) delivers the highest accuracy in the complex task, whereas lighter-weight models (e.g., SLEAP) offer superior speed highlighting a clear trade-off between accuracy and throughput. We also evaluate a recently published foundation model (TopViewMouse-5K) trained on a large top-view mouse dataset to test its generalization. It performs poorly on our tasks at zero-shot, and even when we combine its data with our training set, we observe no consistent benefit. These findings emphasize the importance of context-specific model selection and the need for more diverse training data to create generalizable pose models. By leveraging a general-purpose vision library, researchers can flexibly choose models that best suit their experimental needs. This work illustrates how adopting advanced computer vision frameworks can accelerate behavioral neuroscience and genetics research, paving the way for more scalable, reproducible, and sensitive analysis of animal behavior.

Quantifying feeding behavior with high temporal and spatial precision is critical for understanding how internal state, sensory cues, and neural activity shape food intake and dietary choice. Here, we describe a detailed protocol for performing consumption and dietary choice assays in Drosophila using the flyPAD/optoPAD system. This method enables simultaneous measurement of feeding events across multiple arenas while allowing precise control of gustatory stimuli and optogenetic stimulation. We provide step-by-step instructions for assay food preparation, flyPAD arena setup, data acquisition, and downstream data organization with suggested analyses. This approach is suitable for studying consumption, nutrient preference, learning, and state-dependent modulation of feeding behaviors, and can be readily adapted for optogenetic manipulations and comparative choice assays.

BackgroundHow does neuronal activity change as an animal transitions from being awake to a state of general anesthesia? Previous studies used C. elegans to investigate awake and anesthetized states, emergence from anesthesia, and to establish metrics characterizing how system-wide neuronal dynamics differ under these conditions. This study employs a new technique to image pan-neuronal activity in C. elegans continuously during induction of anesthesia with isoflurane. MethodsC. elegans worms expressing pan-neuronal nuclear RFP and cytosolic GCaMP6s were imaged with light sheet microscopy to measure single cell activity in the majority of neurons in the animals head during induction via isoflurane exposure. Stable concentrations of isoflurane were maintained throughout the experiment by measured flow vaporization of isoflurane into a specially designed gas enclosure compatible with the imaging system. Building on our previous work investigating emergence from anesthesia, we analyzed ensemble neuronal activity, spectrograms of frequency over time, and metrics of information flow between neurons. ResultsInduction of isoflurane anesthesia caused a progressive reduction in neuronal activity over the course of 40 minutes. Spectrograms indicated a loss of bulk signal power across all frequencies, notably in low frequencies too. State Decoupling and Internal Predictability were among the most useful metrics for discriminating the anesthetized state, demonstrating induction kinetics that are the inverse of emergence. However, each animal does not arrive at the anesthetized state at the same time; response times are highly individualized. ConclusionsInformation metrics of neurodynamic activity demonstrate that isoflurane induction results in a gradual increase in neuronal disconnection and disorganization. Thus, at the level of individual neuron connectivity and system dynamics, the induction of anesthesia in C. elegans nematodes is in essence the reverse of emergence. Induction however occurs more rapidly and shows marked variability between individuals. Future genetic studies will show which molecular targets define sensitivity to volatile anesthetics like isoflurane. Summary StatementIsoflurane-induced unconsciousness is a common phenomenon across species. Does the induction of anesthesia arise by distinct state transitions, or through gradual changes in system dynamics when activity is measured at the level of individual neurons?

Inscapes is a low demand abstract animation used as an alternative to eyes open rest in neuroimaging studies, particularly with pediatric and clinical populations prone to head motion. Although prior work has established that functional connectivity patterns during Inscapes closely resemble those during rest, no study has examined whether the two conditions differ in aperiodic neural activity, a broadband feature of the power spectrum linked to excitation/inhibition balance. Here we used magnetoencephalography (MEG) in 54 healthy adults to compare spectrally parameterised aperiodic and periodic measures between eyes open rest and Inscapes viewing (visual component only, without audio). At the sensor level, both the aperiodic exponent and offset were significantly higher during rest than during Inscapes across widespread frontoparietal and occipital distributions in both magnetometers and gradiometers. Source level analyses at both the parcellation and vertex levels largely supported these patterns. The pericalcarine cortex was a notable exception, where both aperiodic measures were higher during Inscapes than during rest, indicating a regionally specific reversal in primary visual cortex. These results demonstrate that Inscapes and eyes open rest produce distinct aperiodic spectral profiles, indicating that the two conditions are not interchangeable for analyses involving broadband spectral dynamics or excitation/inhibition balance estimation.

Recombinant Adeno-associated viruses (AAVs) are widely used for gene delivery in the central nervous system and have become central tools in both gene therapy and basic neuroscience research. However, although AAV serotypes have been extensively characterized in rodent models, their performance in human neurons, particularly those derived from induced pluripotent stem cells (iPSCs), remains poorly characterized. While human iPSC-derived neurons are increasingly used for disease modeling and drug screening, their susceptibility to viral transduction varies and remains difficult to predict. In this study, we systematically evaluated the transduction efficiency and toxicity profiles of 18 wild-type and engineered AAV serotypes across three distinct types of iPSC-derived neurons, relevant to disease modeling and drug discovery: cortical projection neurons, NGN2- induced forebrain-like neurons, and dopaminergic neurons and four doses (1E3, 1E4, 1E5 and 2E5 genome copies per cell). Using automated high-throughput confocal imaging and quantification of reporter gene expression, we identified several serotypes with robust and efficient transduction across all neuronal subtypes. Among these, three serotypes AAV6, AAV6.2 and AAV2.7m8 showed consistently high performance. To assess safety, we quantified cell number and neurite morphology, finding that while high transduction and gene expression correlate with toxicity, sensitivity varied across neuronal subtypes, with NGN2 neurons being most vulnerable and dopaminergic neurons most resilient. Finally, we validated our findings in a more complex 3D model by testing one of the best-performing serotypes, AAV2.7m8, in both whole and dissociated human cerebellar organoids. Together, our results establish a benchmark dataset for AAV performance in human iPSC- derived neurons and provide practical guidance for AAV based gene delivery in human in vitro neural models. This resource will be valuable for both basic research and preclinical applications aiming to manipulate gene expression in human neurons and understanding AAV tropism in disease-relevant cell types.

ObjectiveConvolutional neural networks (CNNs) have shown promise in decoding neural drive from high-density surface electromyography (HD-sEMG) signals. However, the effects of convolutional kernel dimensionality on the generalizability and computational efficiency of CNN-based neural drive decoding remain unclear. This study systematically examined how the dimensionality of convolutional kernels (1D, 2D, and 3D) affects both the generalizability and computational efficiency of CNN-based neural drive decoding. ApproachThree CNN architectures differing only in the dimensionality of their convolutional kernels were implemented to extract temporal (1D), spatial (2D), or spatiotemporal (3D) features from HD-sEMG signals of isometric knee extension, ankle plantarflexion at three intensities. Each CNN was repeatedly trained using subsets of a pooled training dataset with varying sizes. Cross-intensity and cross-muscle generalizability were assessed by the correlation coefficient between neural drive from deep CNN and that from golden standard blind source separation (BSS) algorithms. Computational efficiency was assessed by measuring inference time on both CPU and GPU platforms. Main ResultsAll CNN architectures demonstrated generalizability across contraction intensities and muscles. For cross contraction intensities, the 1D, 2D, and 3D CNNs achieved mean correlation coefficients of 0.986 {+/-} 0.009, 0.987 {+/-} 0.010, and 0.987 {+/-} 0.010, respectively. For cross-muscle generalizability, the corresponding correlation coefficients were 0.961 {+/-} 0.051, 0.965 {+/-} 0.049, and 0.968 {+/-} 0.046. In terms of efficiency, the 3D CNN was the least computationally efficient, with inference times of 4.1 ms per sample on the CPU and 1.2 ms per sample on the GPU. SignificanceThese findings demonstrate that increased CNN architectural complexity does not necessarily yield superior generalizability in neural drive decoding from HD-sEMG signals. The results provide practical guidance for balancing decoding performance and computational efficiency in HD-sEMG-based neural-machine interfaces.

Background and Objectives: We report the first intraoperative deployment of a real-time machine vision system in neurosurgery, derived from our previous anatomical detection work, automatically identifying structures during endoscopic endonasal surgery. Existing systems demonstrate promising performance in offline anatomical recognition, yet so far none have been implemented during live operations. Methods: A real-time anatomy detection model was trained using the YOLOv8 architecture (Ultralytics). Following training completion in the PyTorch environment, the model was exported to ONNX format and further optimized using the NVIDIA TensorRT engine. Deployment was carried out using the NVIDIA Holoscan SDK, the system ran on an NVIDIA Clara AGX developer kit. We used the model for real-time recognition of intraoperative anatomical structures and compared it with the same video labelled manually as reference. Model performance was reported using the average precision at an intersection-over-union threshold of 0.5 (AP50). Furthermore, end-to-end delay from frame acquisition to the display of the annotated output was measured. Results: A mean AP50 of 0.56 was achieved. The model demonstrated reliable detection of the most relevant landmarks in the transsphenoidal corridor. The mean end-to-end latency of the model was 47.81 ms (median 46.57 ms). Conclusion: For the first time, we demonstrate that clinical-grade, real-time machine-vision assistance during neurosurgery is feasible and can provide continuous, automated anatomical guidance from the surgical field. This approach may enhance intraoperative orientation, reduce cognitive load, and offer a powerful tool for surgical training. These findings represent an initial step toward integrating real-time AI support into routine neurosurgical workflows.

Transcranial magnetic stimulation (TMS) is a non-invasive brain stimulation technique widely employed in basic and clinical research. Non-human primates (NHPs) represent translationally valuable models due to their close anatomical and functional similarity to humans. However, significant technical challenges remain in implementing human-like TMS protocols in awake NHPs. Here we developed a non-invasive head- and arm-fixation apparatus that enables reliable stimulation and electromyography recordings in awake NHPs without surgical intervention and validated the apparatus with two TMS protocols in rhesus macaques. First, we implemented an adaptive motor threshold (MT) determination method developed recently for humans, which converged successfully to valid MTs as defined by the International Federation of Clinical Neurophysiology. Second, we measured a robust short-interval intracortical inhibition effect for the first time in awake NHPs. Successful implementation of human TMS protocols in awake NHPs provides proof-of-concept validation of our apparatus, paving the way to bidirectionally translatable, clinically relevant neuromodulation protocols.

Bidirectional interfaces combined with neural de-coding algorithms are essential for closed-loop (CL) neuromodulation, enabling simultaneous neural monitoring and responsive optogenetic stimulation. However, implementing these capabilities in compact wireless headstages for freely moving animals remains challenging, as most existing platforms rely on tethered setups and external processors to execute computationally intensive decoders. This work presents the design and optimization of a neural decoder integrated into a bidirectional wireless system for CL optogenetic experiments in rodents. The proposed platform combines 32-channel electrophysiological recording with neuromorphic feature extraction, dimensionality reduction, and a nonlinear support vector machine (NL-SVM) decoder implemented on a resource-constrained Spartan-6 FPGA. Temporal dynamics are captured using spike-count features and leaky integrators, while principal component analysis (PCA) reduces the feature space to six components, enabling sub-millisecond inference with minimal memory and power requirements. Model size is further reduced using k-means clustering during training to limit the number of support vectors. Decoder performance was validated using datasets from non-human primate and rat motor cortex recordings. The proposed decoder achieved accuracy comparable to convolutional neural networks (R2 =0.85 vs. 0.87) and outperformed Wiener filters (R2 = 0.81) while requiring significantly fewer computational resources. The full system was further demonstrated in vivo through wireless closed-loop optogenetic stimulation in rats, achieving a variance accounted for (VAF) of 0.9148. Overall, this work introduces a versatile, fully self-contained, and resource-efficient platform for real-time untethered closed-loop neuroscience experiments.

Reverse correlation is a widely-used and well-established method for probing latent perceptual representations in which subjects render subjective preference responses to ambiguous stimuli. Stimuli are purposefully designed to have no direct relationship with the target representation (e.g., they are randomly-generated), a property which makes each individual stimulus minimally informative toward reconstructing the target, and often difficult to interpret for subjects. As a result, a large number of stimulus-response pairs must be gathered from a given subject in order for reconstructions to be of sufficient quality, making the task fatiguing. Recent work has demonstrated that the number of trials needed can be substantially reduced using a compressive sensing framework that incorporates the assumption that the target representation can be sparsely represented in some basis into the reconstruction process. Here, we introduce an alternative method that incorporates the sparsity assumption directly into stimulus generation, which holds promise not only for improving efficiency, but also for improving the interpretability of stimuli from subjects perspective. We develop this new method as a mathematical variation of the compressive sensing approach, before conducting one simulation study and two human subjects experiments to assess the benefits of this method to reconstruction quality, sample size efficiency, and subjective interpretability. Results show that sparse stimulus generation improves all three of these areas relative to conventional reverse correlation approaches, and also relative to compressive sensing in most conditions.

Cerebrovascular reactivity (CVR), the ability of cerebral blood vessels to dilate or constrict in response to a vasoactive stimulus, is an important measure of cerebrovascular health. Accurate CVR estimation requires accounting for the time required for the vasoactive stimulus to reach each brain region and the time it takes for local arterioles to modulate cerebral blood flow. The temporal search range used to calculate this spatially varying offset can substantially impact CVR estimates, and the appropriate search range may vary across populations, acquisition protocols, and even brain regions. Here, we present an iterative approach for automatically determining the appropriate maximum shift, using breath-hold fMRI data acquired in a cohort of stroke survivors. This approach selectively expands the delay search range only for voxels with estimated delays at the boundary (i.e., near the minimum or maximum shift) until the estimated delay is no longer constrained or a predefined value is reached. In the context of stroke, this approach significantly increased the number of voxels with statistically significant CVR among those initially at the boundary. It also resulted in CVR polarity reversals in voxels originally at the early-response boundary and amplified negative CVR values in voxels originally at the late-response boundary, suggesting that using an iterative maximum shift can critically impact CVR interpretation. This approach is broadly applicable beyond stroke, but careful parameter tuning is required, as illustrated by our demonstration of the parameter tuning process for a participant with Moyamoya disease. Together, these findings suggest that iterative delay correction allows for improved CVR assessments in clinical populations.