Chia-Lung Hsieh
Institute of Atomic and Molecular Sciences, Academia Sinica
Topic: Machine learning-assisted chromatin imaging in live cell nuclei by label-free interference DYNAMICS imaging
Label-free interference microscopy is a powerful tool for bioimaging, providing optical images associated with the refractive index contrast of the sample. Recent studies showed that the refractive index map of a cell sample contains structural information that is often connected to the fluorescence images of specific cell organelles. However, inferring organelle-specific images solely based on the structural information faces limitations because different cell organelles may exhibit similar refractive index distributions. Here we demonstrate a label-free interference microscope imaging strategy that captures the fluctuation characteristics of the dynamic light scattering signal, referred to as “DYNAMICS imaging”. The DYNAMICS map contains rich temporal information that is closely connected to the molecular fluctuation. Exploiting machine learning, we establish a protocol for high-resolution chromatin imaging in the live cells in a label-free manner.
Keywords: Label-free, interference microscopy, DYNAMICS imaging, cell nucleus, machine learning
Ji-Xin Cheng
Photonics Center, Boston University
Topic: Mid-Infrared Photothermal Microscopy: Principle, Instrumentation, and Applications
Mid-infrared (IR) spectroscopic imaging using inherent vibrational contrast has been broadly used as a powerful analytical tool for sample identification and characterization. However, the low spatial resolution and large water absorption associated with the long IR wavelengths hinder its applications to study subcellular features in living systems. Recently developed mid-infrared photothermal (MIP) microscopy (Figure 1) overcomes these limitations by probing the IR absorption–induced photothermal effect using a visible light [1-6]. MIP microscopy yields sub-micrometer spatial resolution with high spectral fidelity and reduced water background. In this presentation, we overview different mid-infrared photothermal contrast mechanisms and discuss instrumentations for scanning and widefield MIP microscope configurations. We highlight applications from life science to materials.
Katsumasa Fujita
Raman microscopy, Osaka University
Topic:
Coming soon
Jin-Wu Tsai (蔡金吾)
Institute of Brain Science, National Yang Ming Chiao Tung University
Topic: Detection of Neurodegeneration Using Automated Dendritic Spine Identification Based on Convolutional Neural Network
Dendritic spines are tiny protrusions from dendrites to form synapses for receiving neuronal signals. Dendritic spines are progressively lost in neurodegenerative diseases. However, analysis of dendritic spines is still a time-consuming manual task. Here we applied convolution neural network (CNN)-based model for automated detection and classification of dendritic spines. We first captured 3D images of dendritic spines in the mouse cortex by confocal laser scanning microscopy with manually annotated traces of dendrites and dendritic spines. U-Net was used in the training process. The detection accuracy of dendrite and dendritic spines were up to 95.8%. This study showed that CNN-based automated identification approach has a potential to be applied in the field of neurodegeneration and may increase efficiency in analysis of dendritic spines to reduce human bias.
Keywords: neurodegeneration, dendritic spine, convolution neural network, deep learning
Jung-Chi Liao (廖仲麒)
Institute of Atomic and Molecular Sciences, Academia Sinica
Topic: Microscopy-guided subcellular proteomics
Localizing proteins at specific subcellular regions without using target-specific antibodies or fluorescent proteins is challenging. Laser capture microdissection is able to achieve hypothesis-free in-situ protein discovery, but the resolution cannot reach a subcellular level limited by the laser beam size, and the sensitivity is limited by the sensitivity of mass spectrometry, where only high copy-number proteins at a specific dissected region can be identified. Mass spec imaging with localized ionization can achieve a subcellular resolution, but the sensitivity is also highly limited. A major obstacle of in situ proteomic studies versus in situ transcriptomic studies is that proteins lack a PCR-equivalent amplification method to boost sensitivity. Here, in order to achieve hypothesis-free in situ subcellular protein discovery with high sensitivity and high specificity, we integrated methods of microscopy, artificial intelligence, photochemistry, and automation control to perform microscopy-guided targeted opto-biotinylation. Total-sync ultra-content biotinylation of proteins in thousands of fields of view (FOV) with similar morphological features was achieved by FPGA-based mechatronics. Convolutional neural networks-based deep learning is applied to the image to determine the regions of interest (ROIs) in real time. Multifunctional molecules that contain a photochemical warhead (such as Ru(bpy)32+ or benzophenone) and a tagging molecule (such as biotin) were used for protein labeling. The three-step process of imaging-image processing-opto-biotinylation was repeated for thousands of FOVs automatically to enrich biotinylated proteins in similar ROIs enough for avidin bead purification and successive mass spectrometry analysis, effectively beating the limit of protein amplification. With this platform, we were able to validate the technology by showing a >90% specificity and a sensitivity of >1000 species for nuclear proteins. We were also able to identify novel protein players for specific biological problems, including proteins for stress granules positively validated by antibody staining. Together, our total-sync ultra-content microscopic opto-biotinylation method can be applicable to widely diverse cell biology problems to identify novel protein players in the ROIs, enabling hypothesis-free subcellular protein discovery with high sensitivity and specificity.
Light and Neuroscience
Chi-Kuang Sun
Department of Electrical Engineering, National Taiwan University
Distinguished Professor, Department of Electrical Engineering, National Taiwan University
Distinguished Professor of Electrical Engineering and Computer Science
Head of Optical Molecular Imaging Core Laboratory, Molecular Imaging Center, National Taiwan University
Topic: Realtime and Noninvasive Pathological Diagnosis of Diabetic Peripheral Neuropathy by Third-harmonic-generation Imaging of Free Nerve Ending (TIFNE)
Skin biopsy is the current gold standard to provide structural information of free intraepidermal nerve endings (FINEs) for the differential diagnosis of small fiber neuropathy (SFN). The invasive nature, not to mention its technical difficulty and exhausting preparation procedure, make biopsy particularly unfavorable for diabetic peripheral neuropathy (DPN) patients. In this talk, we review our recent work on the development of in vivo label-free third-harmonic-generation imaging of free nerve ending (TIFNE), which is for the first time capable of imaging the 3D structures of unmyelinated FINEs in human skin noninvasively. Its pathological capability was first confirmed by a comparison study with PGP 9.5 immunohistochemistry stained clinical specimens, then followed an in vivo longitudinal spared nerve injury model study. Finally, three in vivo clinical trials were successfully conducted and showed its capability to differentially diagnosing SFN.
Keywords: nerve biopsy, small fiber neuropathy (SFN), diabetic peripheral neuropathy (DPN), third-harmonic-generation imaging of free nerve ending (TIFNE), intraepidermal nerve fiber (IENF) density index
Adam T. Eggebrecht
Biophotonics Research Center
Mallinckrodt Institute of Radiology;
Imaging Sciences Program;
Department of Biomedical Engineering;
Division of Biology and Biomedical Sciences;
Department of Electrical and Systems Engineering.
Washington University School of Medicine
Topic: Developing optical methods for brain mapping at the point-of-care
Mapping of human brain function has revolutionized systems neuroscience. However, traditional functional neuroimaging with positron emission tomography (PET) or functional magnetic resonance imaging (fMRI) cannot be used when applications require portability or are contraindicated because of ionizing radiation (PET) or implanted metal (fMRI). Additionally, current brain mapping methods such as fMRI offer promising sensitivity yet pose significant methodological challenges in studies of awake infants and toddlers due to the loud, constraining environment and extreme susceptibility to motion-induced artifacts. Optical imaging with functional near infrared spectroscopy (fNIRS) has long held promise as a naturalistic neuroimaging technique; however, image quality and reliability have been lacking, due partially to under sampled physiology and poor data-anatomy registration. Recent developments in high density diffuse optical tomography (HD-DOT), a silent, flexible, and scalable technology ideally suited for imaging awake infants and toddlers, have demonstrated dramatically improved image quality over traditional fNIRS methods. In this talk, I will discuss challenges associated with mapping brain function in children in natural settings, recent advancements in developing HD-DOT methods, and applications in childhood development, autism spectrum disorder, and critical care environments.
Ubiquitous Biology & Physiology
Izumi Nishidate
Tokyo University of Agriculture and Technology
Topic: Non-contact physiological measurement using camera-based diffuse reflectance spectroscopy
This paper describes a simple and non-contact measurement of hemoglobin derivatives in including methemoglobin and multiple physiological parameters such as pulse rate (PR), respiratory rate (RR), percutaneous arterial oxygen saturation (SpO2), and tissue oxygen saturation (StO2) using a digital red-green-blue (RGB) camera. Concentrations of hemoglobin derivatives are estimated from videos or images of the skin using a method based on a tissue-like light transport model. The results obtained from human volunteers showed good correlation in of PR, respiratory rate, SpO2 and StO2 between the proposed method and commercially available devices. The results of animal experiments with rats exposed to sodium nitrite demonstrated that the method can evaluate the temporary production of methemoglobin and severe hypoxemia during methemoglobinemia.
Keywords: hemoglobin derivatives, oxygen saturation, pulse rate, respiratory rate, camera-based vital monitoring
En-Te Hwu (胡恩德)
Department of Health Technology, Technical University of Denmark
Topic: Hacking Consumer Electronics for Biomedical Imaging
Trillions of USD have been dedicated to the development of consumer electronic technologies, to produce drones, wireless chargers, spy cameras, and DVD/Blu-ray optical data storage systems. Consequently, these have been perfected into compact, reliable, high-performance, and low-cost devices. This work reviews the complete repurposing (hacking) of consumer electronics for biomedical imaging and sensing applications.
Keywords: hardware hacking, wireless power, lab on disc, optical pickup unit (OPU), atomic force microscopy (AFM)
.
Novel Biomolecular Sensing
Keisuke Goda
Department of Chemistry, University of Tokyo
Topic: Unconventional SERS: metal/plasmon-free and wearable/flexible SERS
I present two types of unconventional surface-enhanced Raman spectroscopy (SERS), namely porous carbon nanowires and gold nanomesh as SERS substrates for metal/plasmon-free and wearable/flexible SERS applications. The porous carbon nanowire substrate provides high Raman signal enhancement (about six orders of magnitude), high reproducibility due to the absence of hot spots, high durability due to no oxidation, and high compatibility to biomolecules due to its fluorescence quenching capability. On the other hand, the gold nanomesh substrate is an easy-to-fabricate, low-cost, ultrathin (less than 50 micrometers), stretchable, and adhesive, allowing wearable/flexible sensing applications in any shape. I show several unique applications of these novel SERS substrates.
Large-tissue and High-speed Imaging
Miya Ishihara
Photo-acoustic imaging, National Defense Medical College
Topic:
Coming soon
Bernhard Baumann
Center for Medical Physics and Biomedical Engineering,
Medical University of Vienna
Topic: Advancing contrast for optical coherence tomography in the eye and brain
Optical coherence tomography (OCT) is an imaging method that provides cross-sectional and three-dimensional images of biological tissue with micrometer-scale resolution and in real-time. OCT image contrast is typically based on the amount of light intensity backscattered from tissue structures. However, additional tissue-specific image contrast can be accessed by measuring the spectral signature or the polarization properties of backscattered light, or by exploiting the high speed of OCT to assess dynamic deformations within the sample. In this presentation, we will discuss advanced contrast mechanisms for OCT as well as their feasibility for imaging of healthy and diseased structures in the eye and brain.
Keywords: optical coherence tomography, biomedical imaging, optical contrast, elastography, polarization
Label-free Microscopy
Laura Waller
Department of Electrical Engineering and Computer Sciences, UC Berkley
Topic: Computational 3D microscopy with scattering samples
Computational imaging involves the joint design of imaging system hardware and software, optimizing across the entire pipeline from acquisition to reconstruction. Computers can replace bulky and expensive optics by solving computational inverse problems, or images can be reconstructed from scattered light. This talk will describe new microscopes that use computational imaging to enable 3D measurements using simple hardware that is easily adoptable and advanced image reconstruction algorithms based on large-scale optimization and learning.
