Abstract:In the new millennium, the brain and neuroscience have taken center-stage in international collaborative efforts. The brain comprises billions of neurons interconnected at trillions of synapses. To decipher its structure and function is one of the boldest projects ever pursued by the scientific community. The development of better imaging technology once again affords powerful tools to meet this grand challenge. Different imaging modalities have been widely used in revealing the complex structural organization and functional dynamics of the brain, including X-ray computed tomography, magnetic resonance imaging, positron emission computed tomography, ultrasound imaging, electron microscopy, and fluorescence microscopy. Among these, only fluorescence microscopy provides high contrast, high specificity, and high spatiotemporal resolution imaging in vivo . Thanks to the recent progress in photonics, laser physics, computer and information science, and nanomaterial science, the century-old optical imaging field is now being revitalizing and is booming. Recent developments in three-photon microscopy enable the optical resolution of single-neuron activity as deep as 2 mm beneath the surface of the cortex, and non-invasively visualization of single-neuron activity through the intact opaque skull. The emergence of photon-efficient super-resolution Hessian structured illumination microscopy allows live cells to be imaged with a spatial resolution 90 nm and an acquisition rate of 564 frames per second, and enables time-lapse super-resolution imaging for over an hour with minimal photo-bleaching. Light-sheet microscopy, on the other hand, is capable of imaging ~100000 neurons in the entire zebrafish brain, at a volumetric imaging rate > 10 Hz. Other innovations such as near-infrared II imaging, photoacoustic tomography, and adaptive optics are also extending the spatial and temporal resolution, imaging depth, and trans-scale volumetric imaging capacity. Another paradigm-shift is to record brain activity in freely-moving and behaving animals, which involves technological innovation in miniaturized microscopy with high spatiotemporal resolution. In this regard, we recently developed a fast, high-resolution, miniaturized two-photon microscope (FHIRM-TPM), with a headpiece weighing only 2.2 g and occupying less than 1 cm3, equipped with a GRIN lens of NA 0.8. Because the fluorescent Ca2+ indicators GFP and GCaMP6 are commonly used in biomedical science, we designed and custom-manufactured a hollow-core photonic crystal fiber to deliver 920-nm femtosecond laser pulses with little dispersion and attenuation. FHIRM-TPM is capable of long-term recording neuronal activity in freely-behaving mice at single-spine and sub-millisecond spatiotemporal resolution (0.64 μm laterally and 3.35 μm axially, 40 Hz at 256 pixel×256 pixel for raster scanning and 10000 Hz for free-line scanning). Future applications of this technology in many behavioral paradigms will help to address many fundamental questions such as spatial and temporal information-processing, learning and memory, decision-making, and social interactions. In summary, with the emphasis on developing more advanced imaging technologies, scientists can directly visualize neuronal activity deeper in the brain, markedly faster with super-resolution, and over many orders of spatiotemporal scales. With technical advances unfolding in multiple fronts (e.g., probes, detectors, modality fusion, and deep-learning assisted imaging), systematic breakthroughs will provide brain scientists and neuroscientists with the ability to gain a holistic view of multi-layered brain activity at the levels of neuron clusters, nuclei, and the circuitry of long-range connectivity. Finally, we envision that high-resolution imaging methods enabling in toto recording of brain activity at single-neuron resolution in small mammals may become a reality within the next decade.

Efficient Implementation of Convolutional Neural Networks in the Data Processing of Two-Photon In Vivo Imaging.

Fast and Lightweight Network Improves Serial Brain Section Stitching

Fast and robust active neuron segmentation in two-photon calcium imaging using spatiotemporal deep learning

A deep-learning approach for online cell identification and trace extraction in functional two-photon calcium imaging

Neurons Identification of Single-Photon Wide-Field Calcium Fluorescent Imaging Data

Penalized matrix decomposition for denoising, compression, and improved demixing of functional imaging data

Classification of A Pressing Movement by Background Removed Single- photon Calcium Image Sequences

Automatic Neuron Detection in Calcium Imaging Data Using Convolutional Networks

Fast, efficient, and accurate neuro-imaging denoising via supervised deep learning

An end-to-end recurrent compressed sensing method to denoise, detect and demix calcium imaging data

Fast, Simple Calcium Imaging Segmentation with Fully Convolutional Networks

Advances In Point Spread Function Engineering For Functional Imaging Of Neural Circuitsin Vivo

A preprocessing toolbox for 2-photon subcellular calcium imaging

In vivo optical imaging of brain activity

Rapid detection of neurons in widefield calcium imaging datasets after training with synthetic data

Single Photon Compressive Imaging with Enhanced Quality Using an Untrained Neural Network.

Efficient and accurate extraction of in vivo calcium signals from microendoscopic video data

High-speed, Large Field-of-view and Deep Imaging with an Adaptive Excitation Source

High-speed low-light in vivo two-photon voltage imaging of large neuronal populations

Recent Advances in Pulse-Coupled Neural Networks with Applications in Image Processing

Ultracompact Deep Neural Network for Ultrafast Optical Property Extraction in Spatial Frequency Domain Imaging (SFDI)