Neural Imaging and Sensing 2025

Two-photon fluorescence microscopy (2PFM) enables activity measurements with calcium indicators deep in the scattering brain with subcellular resolution. However, imaging membrane voltage changes, the most direct measure of neural activity, is more challenging due to its millisecond dynamics. Using a passive optical module based on free-space angular chirp-enhanced delay (FACED), we developed an ultrafast 2PFM system that enabled neural activity imaging in vivo at up to 3,000 frames per second and sub-micron spatial resolution. We have further engineered our system to attain faster imaging rates over 10× larger fields of view. I will discuss our recent results using FACED 2PFM for population imaging of calcium and voltage activity in vivo.

This presentation introduces our recently developed two-photon (2P) fiberscope platform, showcasing its capability for neural imaging in freely-behaving mice. We will highlight several advancements, including innovative miniature optics with cascaded magnification to achieve a large field-of-view without compromising the fiberscope's size and weight, a proactive optoelectrical commutator to sense and eliminate rotational constraints, and a simple apparatus to offset any excessive weight potentially mounted on the mouse’s head in the future. These advances significantly enhance the functionality of the 2P fiberscope for functional neuroimaging in freely-moving rodents. We will also discuss the feasibility for neural imaging studies.

Neurodegenerative disease and neuroscience research require new imaging techniques to address the unmet needs in extracting morphological, molecular, and dynamic information. Central to neurodegenerative studies, tauopathy necessitates non-invasive, sub-cellular 3D imaging of intracellular tau aggregates and their secondary structures. System neuroscience requires 3D, high-speed, large field-of-view imaging to observe neural dynamics in transparent model organisms like jellyfish. To address these challenges, we present advances in computationally enhanced chemical and fluorescent microscopy for 3D in situ imaging of tau fibrils’ secondary structures and high-speed, large field-of-view 3D fluorescent imaging of jellyfish neuroscience models.

Two-Photon Microscopy (TPM) faces limitations due to its tradeoff between temporal frame rate and spatial resolution. While computational imaging techniques stand to significantly extend the limitations of purely optical solutions, they have yet to utilize the spatiotemporal correlations inherent in neural recordings. We propose a new method: Neuroimaging by Oblique Random Acquisition (NORA), which achieves a significant increase in frame rate by compressing spatial signals via an elliptical PSF and randomly subsampling the compressed signal within the FOV. Using simulated data generated from NAOMi configured to simulate the modified optical system, we demonstrate that NORA can successfully recover fluorescence videos with only a tenth of the samples per frame, achieving an effective 10x increase in frame rate.

The miniature multiphoton microscope is a powerful tool for investigating the neuronal basis of complex behaviors by recording brain activity in freely behaving animals. Here we propose multiplexed miniaturized two-photon microscope (M-MINI2P) platforms, which double the imaging speed while maintaining a high spatial resolution. Using M-MINI2P platforms, we conducted high-speed in-vivo neuronal imaging in the mouse cortex. Neuronal signals from multiplexed regions are temporally or computationally demixed and extracted.

Neuroscience has an essential requirement for robust imaging methodologies that operate within the brain’s highly scattering tissue. The high tissue penetrability of ultrasound waves presents untapped and exciting opportunities to meet these challenges. Here we present an overview of recent advances in functional neural imaging based on joint optical-ultrasound imaging. First, I will show how integrating a co-axial ultrasound transducer into a standard multiphoton microscope leads to the observation of a novel gigantic acousto-optic modulation and to ultrasound-modulation assisted multiphoton imaging (UMAMI), and can used to study the mechanisms of transcranial ultrasound neuromodulation. I will then update on recent progress in functional optoacoustic neuro-tomography (FONT), where our recent work demonstrates an optoacoustic signature for state-of-the-art near-infrared genetically encoded calcium indicators. This work highlights emerging new photonic/acoustic hybrid methods for studying the live mouse brain.

We introduce a stretchable, fatigue-resistant hydrogel-based fiber-optic technology for optogenetic modulation of peripheral nerves in mice during sustained locomotion. This system offers significant advancements over rigid devices, with low light attenuation (1.1 dB/cm), high stretchability (75%-150% strain), and excellent fatigue strength (1.4 MPa) against repeated deformations up to 60,000 cycles due to integrated polymeric nanocrystalline domains. The device effectively delivers light pulses to the sciatic nerve, activating hindlimb muscles in six-week voluntary running assays and maintaining functionality after prolonged deformation. Furthermore, it modulates sensory afferents, demonstrated by reducing nociceptive pain in TRPV1::NpHR mice over two months. This hydrogel fiber-optic technology provides a flexible, durable solution for targeting peripheral neural circuits under continuous deformation, facilitating studies on motor recovery and pain processing.

GRIN lens has been widely used for deep brain recording applications. However, the imaging field of view through GRIN lenses is typically very small, limited by the strong inherent aberration of GRIN lenses. Moreover, the limited working distance of GRIN lenses confines the imaging to tissue immediately adjacent to the lens facet. Thus, the tissue health near the facet determines the success rate of the measurement. To enable large-volume calcium imaging through 0.5 mm GRIN lenses with a high success rate, we developed an integrated solution that allows better tissue health, greater success rate, and large imaging volume.

Mental health issues are a growing concern in our society. Its impact on individual well-being and health cannot be understated. In response, scientists have designed tools to gauge mental stress in its early phases. Electroencephalogram (EEG) signals can provide comprehensive insight into mental states and conditions. In this project, we propose to use Variational-Autoencoders that extract important features and reduce noise from EEG signals. Next, we will build and compare different machine learning and deep learning models for EEG sentiment analysis. The different Deep Learning models will be trained on EEG data to find patterns of brainwave activity associated with different emotions. After preprocessing, normalization and noise removal using Variational Autoencoders, we implemented and compared different models. To achieve the best results, we used different optimizers, loss functions and experimented with early stopping and regularization layers to prevent overfitting. SVM results show 89% accuracy, while the CNNs and RNNs results show 96% accuracy.

Intravital imaging has presented a powerful tool for tracking cellular function and structural dynamics in the real environment of a complex multicellular organism. However, most commercial microscopy systems are limited by optical design, allowing only long-term live imaging of transparent samples, such as Drosophila embryo, zebrafish embryo, and suffering from slow image acquisition speeds. Consequently, they can only generate two-dimensional time-lapse images at selected optical planes. Here, we introduce a high-resolution Bessel beam based vertical digital scanning light sheet microscopy (V-SPIM) for real-time imaging of the adult fruit fly brain. Compared to traditional multiphoton imaging or multi-point confocal microscopy, Bessel beam based light sheet microscope exhibits lower photodamage and faster acquisition volume rates (V.R.) higher than 1Hz (with proper exposure time (E.T.) and signal to background ratio (SNR)) for each volume(330×330×162𝜇𝑚), allowing us to perform long-term volumetric time-lapse functional imaging in adult animals. This novel approach allows us to visualize the olfactory coding of various odors in a multitude of cells within a single fly brain.

Functional connectivity (FC) analysis has gained prominence in human neuroimaging studies utilizing functional magnetic resonance imaging (fMRI) as a tool in mapping spontaneous brain activity. This method investigates the functional relationships between neural activity and has been used extensively in clinical populations as well as in animal models to probe healthy function and disease-related dysfunction. As FC methods advance; a gap continues to grow between human fcMRI and the detailed genetic and molecular techniques common in mouse models. In this study, we aim to bridge this gap by leveraging recent advancements in large field-of-view two-photon microscopy and genetically encoded calcium indicators (GECIs). Together these tools enable examining FC at both the neuronal level, and mesoscopic level in mouse models. Bilateral connections typically indicate strong connectivity between homotopic contralateral regions that share functionality. While the process takes two objectives, this represents a useful step towards integrating neuron-level analysis with mesoscopic systems-level approach of FC.

Knowledge about the neuronal dynamics and the projectome are both essential for understanding how the neuronal network functions in concert. However, it remains challenging to obtain the neural activity and the brain-wide projectome for the same neurons, especially for neurons in subcortical brain regions. Here, by combining in vivo microscopy and high-definition fluorescence micro-optical sectioning tomography（HD-fMOST）, we have developed strategies for mapping the brain-wide projectome of functionally relevant neurons in the somatosensory cortex, the dorsal hippocampus, and the substantia nigra pars compacta. More importantly, we also developed a strategy to achieve acquiring the neural dynamic and brain-wide projectome of the molecularly defined neuronal subtype. The strategies developed in this study solved the essential problem of linking brain-wide projectome to neuronal dynamics for neurons in subcortical structures and provided valuable approaches for understanding how the brain is functionally organized via intricate connectivity patterns.

The myenteric plexus (MP) is a crucial component of the enteric nervous system (ENS), capable of autonomously controlling the motility of the gastrointestinal tract. The direct connection between the central nervous system (CNS) and ENS via the vagus nerve (VN) plays a significant role in maintaining digestive function. Due to the challenges associated with real-time in vivo recording of the ENS, our understanding of the CNS's real-time influence on ENS activity is limited. We have developed a method that allows for real-time in vivo recording of the ENS, enabling stable recording of the real-time effects of vagal activation on the gastric MP.

This study introduces an innovative method for integrating optical wireless communication (OWC) and sensing between brain implants and external interfaces, utilizing shared optical resources. By employing a modulated retroreflector (MRR), the system significantly lowers implant energy consumption, enabling more efficient communication. Additionally, simultaneous sensing uses the same optical resources and applies diffuse optical tomography (DOT) to monitor tissue responses, such as inflammation around the implant. Finally, regression models in the external system predict OWC performance and troubleshoot issues when biological responses hinder the communication link. This integrated approach enhances brain-computer interfaces, promising improved patient outcomes and management.

As an emerging neuroscience model, jellyfish provide unique insights into the nervous system's structure, function, and evolution due to their preserved ancient decentralized nervous system, associated complex behaviors, and remarkable regenerative abilities. Recently established transgenic jellyfish offer a powerful model that is simple, genetically tractable, and optically transparent. However, existing imaging methods lack the capability to capture 3D neural activities in vivo with the necessary high speed and large field of view. To address this, we develop a computational 3D fluorescent microscope that enables video-rate, dual-color in vivo 3D neuron imaging of jellyfish with a millimeter-level field of view.

Characterizing the three-dimensional (3D) structure of the human brain is complex. The use of light sheet fluorescence microscopy (LSFM) has enabled significant progress, allowing the study of brain architecture with unprecedented resolution. We combined LSFM, tissue clearing protocols, and advanced bioimage analysis to create detailed maps of neurons and fiber orientation. Using the SHORT method and a double-view LSFM microscope, along with automated tools BCFindv2 and Foa3D, we analyzed Broca’s area and the brainstem. This work demonstrates the feasibility of fast, high-resolution 3D reconstructions of the human brain.

Neuroinflammation in Alzheimer’s disease alters astrocytic function and morphology, and consequently, may affect their ability to regulate cerebral blood flow. Here, we measured astrocyte Ca2+ and cerebral hemodynamics in the cortex of awake, 12-month-old female APPswe:PS1dE9 mice over the course of a two-week inflammatory threat using two-photon microscopy. Astrocyte Ca2+ and brain hemodynamics at resting-state and during somatosensory stimulation was assessed. We showed inflammation reduced resting-state astrocyte Ca2+ activity. Inflammation-induced reduction in arteriole dilation during sustained sensory stimulation is paired with the reduction in astrocyte endfeet Ca2+, suggesting the correlation between impaired astrocyte Ca2+ and reduced arteriole dilation.

Deep brain stimulation (DBS) is an invasive treatment for patients suffering from Parkinson’s disease. Generally, there are two categories of DBS: open-loop DBS and closed-loop DBS. Open-loop DBS uses constant current for neural stimulation and has no feedback signal. Thus, it cannot immediately adjust stimulation conditions to optimize the stimulation effect. In contrast, closed-loop DBS can dynamically adjust the stimulation condition based on the sensed signal to improve the therapeutic effect. Furthermore, constructing a feedback model to optimize stimulation is an important issue. The electrical impedance model characterizes how the impedance varies with frequency and is constituted by resistors, capacitors, and inductors. This model can predict how the brain responds to stimulation and is used as a feedback model for closed-loop DBS. Thus, this study proposes an approach to construct the electrical impedance model of the subthalamic nucleus in a swine based on the closed-loop DBS.

Wearable high-density diffuse optical tomography (HD-DOT) systems often use PIN photodiodes rather than avalanche photodiodes (APDs), as their low bias voltages allow a smaller form factor, but make a tradeoff with noise performance as they lack the intrinsic gain of APDs. The use of silicon photomultipliers (SiPMs) in neuroimaging systems provides the opportunity to achieve the noise characteristics of fiber systems in the small format required for wearable HD-DOT systems. We integrate custom SiPM-based source/detector modules into an imaging cap, showing these improved noise characteristics extend the range of high-SNR measurements, increasing overall system channel count.

Despite the success of genetically encoded functional indicators, most of the measurement has been limited to the mouse neocortex. Deep brain region is hard to access due to scattering. Currently, the common method remains the GRIN lens-based imaging. However, a major drawback of GRIN lens-based approaches is that the imaging volume is tiny. Towards, large volume deep brain imaging, Clear Optically Matched Panoramic Access Channel Technique (COMPACT) has been developed, which allows panoramic imaging through thin wall glass capillaries. Over the past few years, we have significantly advanced the technology to achieve large-volume mesoscopic calcium recording through 0.5 mm probe.

This study presents an approach for functional neural imaging and sensing using fluorescent nanosensors designed to detect potassium (K+) flux in living systems. The nanoprobes exhibit a change in fluorescence that is dependent on local K+ concentration to facilitate real-time sensing of ionic gradients crucial for understanding neuronal activity and ion channel dynamics. We demonstrated K+ sensing with our nanoprobes by electrically stimulating murine brain slices to induce ionic flux. This method offers a useful tool for studying ion-dependent processes in real-time, providing new insights into the complex functioning of neural networks.

In neuro-surgery, when opening up the dura mater, a phenomenon called "brain shift" occurs: Due to the internal pressure of the tissue, the brain will expand by several millimeters into the situs opening. Due to this expansion, neuro-navigation surgical guidance data obtained pre-operatively using magnetic resonance imaging or computer tomography no longer accurately represent a patient's brain's geometry during surgery. It has been shown that transdural OCT imaging can picture anatomy in the subdural space. However, this used to happen only in 2D cross-sections, which made it difficult to precisely localize anatomical landmarks in 3D space. We demonstrate high-resolution human in-vivo transdural 3D volume image acquisition in the subdural space using our microscope-integrated MHz optical coherence tomography system, with a 12mm×7mm scan done in under 2 seconds. Our proposed method, which involves acquiring transdural 3D volumes prior to opening the dura mater and then after, offers a promising solution. By obtaining data on the mechanical deformations, we can potentially enhance the accuracy of neuro-navigation surgical guidance by introducing geometrical corrections.

We introduce a rapid dual-view projection imaging technique with two-photon glutamate uncaging capabilities. This method significantly enhances volumetric imaging rates for neural studies over traditional light-sheet imaging. Furthermore, we integrated an independent laser scanning module for two-photon uncaging, allowing for simultaneous synaptic resolution stimulation and 100 Hz volumetric imaging of neural activity in deep tissue. Additionally, we provide imaging results from mouse brain slices undergoing two-photon glutamate uncaging.

We present a novel short-line excitation strategy for two-photon microscopy to enhance imaging speed and efficiency in neuronal activity recording. This method, implemented in both benchtop and miniaturized two-photon microscopes, significantly increases the frame rates by sampling multiple points simultaneously and reducing the number of scanned rows. Using this method, we demonstrated high-speed imaging of neuronal activity in both head-fixed and freely-moving mice. This advancement holds great promise for studying the brain activity.

Nowadays, optical microscopy with high resolution, contrast, and sensitivity has been considered as an indispensability tool for many applications, including fundamental research, clinical application, and industrial measurement. Phase microscopy, as a very important branch, takes the natural property such as refractive index instead of the extra agents as contrast for label-free imaging of biological tissues. In this work, we propose a high-speed swept-source-based common-path quantitative phase microscopy with high-sensitivity, non-label, and three-dimensional imaging features.

The hippocampus is a crucial brain region involved in functions such as memory, learning, spatial navigation, and emotional regulation. Neuronal diseases such as Alzheimer's, depression, epilepsy, and schizophrenia often show a reduction in hippocampal volume. This study aims to propose a novel method for measuring hippocampal volume and observing its structure using Multiscale Serial Optical Coherence Microscopy (MS-OCM). Additionally, to investigate the key biomarkers for neurodegenerative disease such as reactive astrocyte, we observed identical sample with confocal microscopy. Eventually, we could quantify key biomarkers through multiple modalities and multiple scale levels in volumetric manners.

Visual semantic mapping connects object category labels to language processing across extensive cortical regions. While fMRI is effective for semantic encoding (mapping) and decoding, it is costly and unsuitable for applications in natural environments. High-density diffuse optical tomography (HD-DOT) offers a cost-effective alternative for imaging in naturalistic settings. We developed a semantic encoding model using very high-density DOT (VHD-DOT). Participants (N=3) underwent three 89-minute VHD-DOT imaging sessions, including naturalistic movie clips (120 minutes training and 90 minutes testing data). Semantic maps were derived for single object categories using a GLM and a full set of categories using a voxelwise encoding model. VHD-DOT demonstrated high repeatability and sufficient signal quality across imaging sessions, as well as reproducible single-category and rich multiple-category semantic maps. This study confirms the feasibility of VHD-DOT for semantic mapping, building towards the development of brain-computer interfaces for individuals with language difficulties, like stroke-induced aphasia, in natural environments.

Deep brain stimulation (DBS) treats symptoms of Parkinson's disease, such as tremors, muscle rigidity, and bradykinesia. In closed-loop DBS, neural signals are synchronously sensed alongside stimulation signals to adjust the stimulation for optimal therapeutic effect. This study introduces an optimized DBS method using electrical impedance models. These models reflect the brain's response to electrical signals, enhancing signal accuracy. Using Parkinson's disease as an example, local field potentials (LFP) in the swine subthalamic nucleus (STN) are sensed, and the beta band is extracted as a biomarker for Parkinson's disease. Sensory data is converted into stimulation currents using electrical impedance models. The power consumption and effectiveness of the stimulation using electrical impedance models are compared to those of the stimulation using the traditional threshold method. Eventually, the utility of the electrical impedance models for closed-loop DBS parameter feedback was demonstrated, and the beneficial effects on Parkinson's disease in swine were highlighted.

The advancement of optical neuroimaging requires improvements in size/weight without sacrificing system performance. A modular wearable imaging system is redesigned with a focus on minimizing detector noise, resulting in the development of a system with significant size and weight improvements while maintaining key detector characteristics. System bench specifications are evaluated, showing a noise-equivalent-power of 91fW/√Hz, and imaging, including retinotopy, is performed in-vivo, showing contralateral activation on the visual cortex. Modules are arranged for high-density (13-mm first nearest-neighbor spacing) whole-head coverage, providing a path toward high-quality optical neuroimaging in naturalistic environments.