ABSTRACT
Plastics are now omnipresent in our daily lives. The existence of microplastics (1 µm to 5 mm in length) and possibly even nanoplastics (<1 µm) has recently raised health concerns. In particular, nanoplastics are believed to be more toxic since their smaller size renders them much more amenable, compared to microplastics, to enter the human body. However, detecting nanoplastics imposes tremendous analytical challenges on both the nano-level sensitivity and the plastic-identifying specificity, leading to a knowledge gap in this mysterious nanoworld surrounding us. To address these challenges, we developed a hyperspectral stimulated Raman scattering (SRS) imaging platform with an automated plastic identification algorithm that allows micro-nano plastic analysis at the single-particle level with high chemical specificity and throughput. We first validated the sensitivity enhancement of the narrow band of SRS to enable high-speed single nanoplastic detection below 100 nm. We then devised a data-driven spectral matching algorithm to address spectral identification challenges imposed by sensitive narrow-band hyperspectral imaging and achieve robust determination of common plastic polymers. With the established technique, we studied the micro-nano plastics from bottled water as a model system. We successfully detected and identified nanoplastics from major plastic types. Micro-nano plastics concentrations were estimated to be about 2.4 ± 1.3 × 105 particles per liter of bottled water, about 90% of which are nanoplastics. This is orders of magnitude more than the microplastic abundance reported previously in bottled water. High-throughput single-particle counting revealed extraordinary particle heterogeneity and nonorthogonality between plastic composition and morphologies; the resulting multidimensional profiling sheds light on the science of nanoplastics.
Subject(s)
Drinking Water , Microscopy , Humans , Microplastics , Plastics , AlgorithmsABSTRACT
Tuberculosis (TB) is the world's deadliest infectious disease, with over 1.5 million deaths and 10 million new cases reported anually. The causative organism Mycobacterium tuberculosis (Mtb) can take nearly 40 d to culture, a required step to determine the pathogen's antibiotic susceptibility. Both rapid identification and rapid antibiotic susceptibility testing of Mtb are essential for effective patient treatment and combating antimicrobial resistance. Here, we demonstrate a rapid, culture-free, and antibiotic incubation-free drug susceptibility test for TB using Raman spectroscopy and machine learning. We collect few-to-single-cell Raman spectra from over 25,000 cells of the Mtb complex strain Bacillus Calmette-Guérin (BCG) resistant to one of the four mainstay anti-TB drugs, isoniazid, rifampicin, moxifloxacin, and amikacin, as well as a pan-susceptible wildtype strain. By training a neural network on this data, we classify the antibiotic resistance profile of each strain, both on dried samples and on patient sputum samples. On dried samples, we achieve >98% resistant versus susceptible classification accuracy across all five BCG strains. In patient sputum samples, we achieve ~79% average classification accuracy. We develop a feature recognition algorithm in order to verify that our machine learning model is using biologically relevant spectral features to assess the resistance profiles of our mycobacterial strains. Finally, we demonstrate how this approach can be deployed in resource-limited settings by developing a low-cost, portable Raman microscope that costs <$5,000. We show how this instrument and our machine learning model enable combined microscopy and spectroscopy for accurate few-to-single-cell drug susceptibility testing of BCG.
Subject(s)
Antitubercular Agents , Machine Learning , Microbial Sensitivity Tests , Mycobacterium tuberculosis , Spectrum Analysis, Raman , Spectrum Analysis, Raman/methods , Mycobacterium tuberculosis/drug effects , Humans , Microbial Sensitivity Tests/methods , Antitubercular Agents/pharmacology , Drug Resistance, Bacterial , Tuberculosis, Multidrug-Resistant/drug therapy , Tuberculosis, Multidrug-Resistant/microbiology , Tuberculosis, Multidrug-Resistant/diagnosis , Tuberculosis/drug therapy , Tuberculosis/microbiology , Isoniazid/pharmacologyABSTRACT
Chemical transformations near plasmonic metals have attracted increasing attention in the past few years. Specifically, reactions occurring within plasmonic nanojunctions that can be detected via surface and tip-enhanced Raman (SER and TER) scattering were the focus of numerous reports. In this context, even though the transition between localized and nonlocal (quantum) plasmons at nanojunctions is documented, its implications on plasmonic chemistry remain poorly understood. We explore the latter through AFM-TER-current measurements. We use two molecules: i) 4-mercaptobenzonitrile (MBN) that reports on the (non)local fields and ii) 4-nitrothiophenol (NTP) that features defined signatures of its neutral/anionic forms and dimer product, 4,4'-dimercaptoazobenzene (DMAB). The transition from classical to quantum plasmons is established through our optical measurements: It is marked by molecular charging and optical rectification. Simultaneously recorded force and current measurements support our assignments. In the case of NTP, we observe the parent and DMAB product beneath the probe in the classical regime. Further reducing the gap leads to the collapse of DMAB to form NTP anions. The process is reversible: Anions subsequently recombine into DMAB. Our results have significant implications for AFM-based TER measurements and their analysis, beyond the scope of this work. In effect, when precise control over the junction is not possible (e.g., in SER and ambient TER), both classical and quantum plasmons need to be considered in the analysis of plasmonic reactions.
ABSTRACT
The interplay of charge, spin, lattice, and orbital degrees of freedom in correlated materials often leads to rich and exotic properties. Recent studies have brought new perspectives to bosonic collective excitations in correlated materials. For example, inelastic neutron scattering experiments revealed non-trivial band topology for magnons and spin-orbit excitons (SOEs) in a quantum magnet CoTiO3 (CTO). Here, we report phonon properties resulting from a combination of strong spin-orbit coupling, large crystal field splitting, and trigonal distortion in CTO. Specifically, the interaction between SOEs and phonons endows chirality to two [Formula: see text] phonon modes and leads to large phonon magnetic moments observed in magneto-Raman spectra. The remarkably strong magneto-phononic effect originates from the hybridization of SOEs and phonons due to their close energy proximity. While chiral phonons have been associated with electronic topology in some materials, our work suggests opportunities may arise by exploring chiral phonons coupled to topological bosons.
ABSTRACT
Accelerating the measurement for discrimination of samples, such as classification of cell phenotype, is crucial when faced with significant time and cost constraints. Spontaneous Raman microscopy offers label-free, rich chemical information but suffers from long acquisition time due to extremely small scattering cross-sections. One possible approach to accelerate the measurement is by measuring necessary parts with a suitable number of illumination points. However, how to design these points during measurement remains a challenge. To address this, we developed an imaging technique based on a reinforcement learning in machine learning (ML). This ML approach adaptively feeds back "optimal" illumination pattern during the measurement to detect the existence of specific characteristics of interest, allowing faster measurements while guaranteeing discrimination accuracy. Using a set of Raman images of human follicular thyroid and follicular thyroid carcinoma cells, we showed that our technique requires 3,333 to 31,683 times smaller number of illuminations for discriminating the phenotypes than raster scanning. To quantitatively evaluate the number of illuminations depending on the requisite discrimination accuracy, we prepared a set of polymer bead mixture samples to model anomalous and normal tissues. We then applied a home-built programmable-illumination microscope equipped with our algorithm, and confirmed that the system can discriminate the sample conditions with 104 to 4,350 times smaller number of illuminations compared to standard point illumination Raman microscopy. The proposed algorithm can be applied to other types of microscopy that can control measurement condition on the fly, offering an approach for the acceleration of accurate measurements in various applications including medical diagnosis.
Subject(s)
Microscopy , Spectrum Analysis, Raman , Humans , Microscopy/methods , Spectrum Analysis, Raman/methods , Thyroid Gland , Nonlinear Optical Microscopy , Machine LearningABSTRACT
Nanomedicine has emerged as a revolutionary strategy of drug delivery. However, fundamentals of the nano-neuro interaction are elusive. In particular, whether nanocarriers can cross the blood-brain barrier (BBB) and release the drug cargo inside the brain, a basic process depicted in numerous books and reviews, remains controversial. Here, we develop an optical method, based on stimulated Raman scattering, for imaging nanocarriers in tissues. Our method achieves a suite of capabilities-single-particle sensitivity, chemical specificity, and particle counting capability. With this method, we visualize individual intact nanocarriers crossing the BBB of mouse brains and quantify the absolute number by particle counting. The fate of nanocarriers after crossing the BBB shows remarkable heterogeneity across multiple scales. With a mouse model of aging, we find that blood-brain transport of nanocarriers decreases with age substantially. This technology would facilitate development of effective therapeutics for brain diseases and clinical translation of nanocarrier-based treatment in general.
Subject(s)
Brain Diseases , Nanomedicine , Animals , Mice , Brain/diagnostic imaging , Blood-Brain Barrier/diagnostic imaging , AgingABSTRACT
Bestrhodopsins constitute a class of light-regulated pentameric ion channels that consist of one or two rhodopsins in tandem fused with bestrophin ion channel domains. Here, we report on the isomerization dynamics in the rhodopsin tandem domains of Phaeocystis antarctica bestrhodopsin, which binds all-trans retinal Schiff-base (RSB) absorbing at 661 nm and, upon illumination, converts to the meta-stable P540 state with an unusual 11-cis RSB. The primary photoproduct P682 corresponds to a mixture of highly distorted 11-cis and 13-cis RSB directly formed from the excited state in 1.4 ps. P673 evolves from P682 in 500 ps and contains highly distorted 13-cis RSB, indicating that the 11-cis fraction in P682 converts to 13-cis. Next, P673 establishes an equilibrium with P595 in 1.2 µs, during which RSB converts to 11-cis and then further proceeds to P560 in 48 µs and P540 in 1.0 ms while remaining 11-cis. Hence, extensive isomeric switching occurs on the early ground state potential energy surface (PES) on the hundreds of ps to µs timescale before finally settling on a metastable 11-cis photoproduct. We propose that P682 and P673 are trapped high up on the ground-state PES after passing through either of two closely located conical intersections that result in 11-cis and 13-cis RSB. Co-rotation of C11=C12 and C13=C14 bonds results in a constricted conformational landscape that allows thermal switching between 11-cis and 13-cis species of highly strained RSB chromophores. Protein relaxation may release RSB strain, allowing it to evolve to a stable 11-cis isomeric configuration in microseconds.
Subject(s)
Diterpenes , Retinaldehyde , Rhodopsin , Isomerism , Protein Conformation , Rhodopsin/metabolism , Retinaldehyde/chemistryABSTRACT
Plasmonic materials are promising photocatalysts as they are well suited to convert light into hot carriers and heat. Hot electron transfer is suggested as the driving force in many plasmon-driven reactions. However, to date, there are no direct molecular measures of the rate and yield of plasmon-to-molecule electron transfer or energy of these electrons on the timescale of plasmon decay. Here, we use ultrafast and spectroelectrochemical surface-enhanced Raman spectroscopy to quantify electron transfer from a plasmonic substrate to adsorbed methyl viologen molecules. We observe a reduction yield of 2.4 to 3.5% on the picosecond timescale, with plasmon-induced potentials ranging from [Formula: see text]3.1 to [Formula: see text]4.5 mV. Excitingly, some of these reduced species are stabilized and persist for tens of minutes. This work provides concrete metrics toward optimizing material-molecule interactions for efficient plasmon-driven photocatalysis.
ABSTRACT
Development of a simple, label-free screening technique capable of precisely and directly sensing interaction-in-solution over a size range from small molecules to large proteins such as antibodies could offer an important tool for researchers and pharmaceutical companies in the field of drug development. In this work, we present a thermostable Raman interaction profiling (TRIP) technique that facilitates low-concentration and low-dose screening of binding between protein and ligand in physiologically relevant conditions. TRIP was applied to eight protein-ligand systems, and produced reproducible high-resolution Raman measurements, which were analyzed by principal component analysis. TRIP was able to resolve time-depending binding between 2,4-dinitrophenol and transthyretin, and analyze biologically relevant SARS-CoV-2 spike-antibody interactions. Mixtures of the spike receptor-binding domain with neutralizing, nonbinding, or binding but nonneutralizing antibodies revealed distinct and reproducible Raman signals. TRIP holds promise for the future developments of high-throughput drug screening and real-time binding measurements between protein and drug.
Subject(s)
COVID-19 , Microscopy , Humans , SARS-CoV-2 , Drug Evaluation, Preclinical , Ligands , Antibodies, Viral , Drug Interactions , Spike Glycoprotein, Coronavirus/metabolism , Antibodies, NeutralizingABSTRACT
Aerosol microdroplets as microreactors for many important atmospheric reactions are ubiquitous in the atmosphere. pH largely regulates the chemical processes within them; however, how pH and chemical species spatially distribute within an atmospheric microdroplet is still under intense debate. The challenge is to measure pH distribution within a tiny volume without affecting the chemical species distribution. We demonstrate a method based on stimulated Raman scattering microscopy to visualize the three-dimensional pH distribution inside single microdroplets of varying sizes. We find that the surface of all microdroplets is more acidic, and a monotonic trend of pH decreasing is observed in the 2.9-µm aerosol microdroplet from center to edge, which is well supported by molecular dynamics simulation. However, bigger cloud microdroplet differs from small aerosol for pH distribution. This size-dependent pH distribution in microdroplets can be related to the surface-to-volume ratio. This work presents noncontact measurement and chemical imaging of pH distribution in microdroplets, filling the gap in our understanding of spatial pH in atmospheric aerosol.
ABSTRACT
Water is a ubiquitous and vital component of living systems. Hydration, which is the interaction between water and intracellular biomolecules, plays an important role in cellular processes. However, it is technically challenging to study water structure within cells directly. Here, we demonstrate the utility and power of the water bend-libration combination band as a unique Raman spectral imaging probe of cellular hydration. Hydration maps reveal distinct water environments within subcellular compartments (e.g., nucleolus and lipid droplet) due to the spectral sensitivity of this coupled vibrational band. Spectroscopic studies using the water bend-libration are broadly applicable, offering the potential to capture the chemical complexity of hydration in numerous systems.
Subject(s)
Spectrum Analysis, Raman , Water , Water/chemistry , Spectrum AnalysisABSTRACT
Aldehydes fixation was accidentally discovered in the early 20th century and soon became a widely adopted practice in the histological field, due to an excellent staining enhancement in tissues imaging. However, the fixation process itself entails cell proteins denaturation and crosslinking. The possible presence of artifacts, that depends on the specific system under observation, must therefore be considered to avoid data misinterpretation. This contribution takes advantage of scanning electron assisted-dielectric microscopy (SE-ADM) and Raman 2D imaging to reveal the possible presence and the nature of artifacts in unstained, and paraformldehyde, PFA, fixed MNT-1 cells. The high resolution of the innovative SE-ADM technique allowed the identification of globular protein clusters in the cell cytoplasm, formed after protein denaturation and crosslinking. Concurrently, SE-ADM images showed a preferential melanosome adsorption on the cluster's outer surface. The micron-sized aggregates were discernible in Raman 2D images, as the melanosomes signal, extracted through 2D principal component analysis, unequivocally mapped their location and distribution within the cells, appearing randomly distributed in the cytoplasm. Protein clusters were not observed in living MNT-1 cells. In this case, mature melanosomes accumulate preferentially at the cell periphery and are more closely packed than in fixed cells. Our results show that, although PFA does not affect the melanin structure, it disrupts melanosome distribution within the cells. Proteins secondary structure, conversely, is partially lost, as shown by the Raman signals related to α-helix, ß-sheets, and specific amino acids that significantly decrease after the PFA treatment.
Subject(s)
Melanins , Melanosomes , Microscopy, Electron, Scanning , Melanosomes/metabolism , Melanins/metabolismABSTRACT
The tumor microenvironment (TME) is a dynamic pseudoorgan that shapes the development and progression of cancers. It is a complex ecosystem shaped by interactions between tumor and stromal cells. Although the traditional focus has been on the paracrine communication mediated by protein messengers, recent attention has turned to the metabolic secretome in tumors. Metabolic enzymes, together with exchanged substrates and products, have emerged as potential biomarkers and therapeutic targets. However, traditional techniques for profiling secreted metabolites in complex cellular contexts are limited. Surface-enhanced Raman scattering (SERS) has emerged as a promising alternative due to its nontargeted nature and simplicity of operation. Although SERS has demonstrated its potential for detecting metabolites in biological settings, its application in deciphering metabolic interactions within multicellular systems like the TME remains underexplored. In this study, we introduce a SERS-based strategy to investigate the secreted purine metabolites of tumor cells lacking methylthioadenosine phosphorylase (MTAP), a common genetic event associated with poor prognosis in various cancers. Our SERS analysis reveals that MTAP-deficient cancer cells selectively produce methylthioadenosine (MTA), which is taken up and metabolized by fibroblasts. Fibroblasts exposed to MTA exhibit: i) molecular reprogramming compatible with cancer aggressiveness, ii) a significant production of purine derivatives that could be readily recycled by cancer cells, and iii) the capacity to secrete purine derivatives that induce macrophage polarization. Our study supports the potential of SERS for cancer metabolism research and reveals an unprecedented paracrine crosstalk that explains TME reprogramming in MTAP-deleted cancers.
Subject(s)
Ecosystem , Neoplasms , Humans , Neoplasms/drug therapy , Purines/metabolism , Purine-Nucleoside Phosphorylase/genetics , Tumor MicroenvironmentABSTRACT
Establishing the fundamental chemical principles that govern molecular electronic quantum decoherence has remained an outstanding challenge. Fundamental questions such as how solvent and intramolecular vibrations or chemical functionalization contribute to the decoherence remain unanswered and are beyond the reach of state-of-the-art theoretical and experimental approaches. Here we address this challenge by developing a strategy to isolate electronic decoherence pathways for molecular chromophores immersed in condensed phase environments that enables elucidating how electronic quantum coherence is lost. For this, we first identify resonance Raman spectroscopy as a general experimental method to reconstruct molecular spectral densities with full chemical complexity at room temperature, in solvent, and for fluorescent and non-fluorescent molecules. We then show how to quantitatively capture the decoherence dynamics from the spectral density and identify decoherence pathways by decomposing the overall coherence loss into contributions due to individual molecular vibrations and solvent modes. We illustrate the utility of the strategy by analyzing the electronic decoherence pathways of the DNA base thymine in water. Its electronic coherences decay in [Formula: see text]30 fs. The early-time decoherence is determined by intramolecular vibrations while the overall decay by solvent. Chemical substitution of thymine modulates the decoherence with hydrogen-bond interactions of the thymine ring with water leading to the fastest decoherence. Increasing temperature leads to faster decoherence as it enhances the importance of solvent contributions but leaves the early-time decoherence dynamics intact. The developed strategy opens key opportunities to establish the connection between molecular structure and quantum decoherence as needed to develop chemical strategies to rationally modulate it.
ABSTRACT
The large-scale implementation of renewable energy systems necessitates the development of energy storage solutions to effectively manage imbalances between energy supply and demand. Herein, we investigate such a scalable material solution for energy storage in supercapacitors constructed from readily available material precursors that can be locally sourced from virtually anywhere on the planet, namely cement, water, and carbon black. We characterize our carbon-cement electrodes by combining correlative EDS-Raman spectroscopy with capacitance measurements derived from cyclic voltammetry and galvanostatic charge-discharge experiments using integer and fractional derivatives to correct for rate and current intensity effects. Texture analysis reveals that the hydration reactions of cement in the presence of carbon generate a fractal-like electron-conducting carbon network that permeates the load-bearing cement-based matrix. The energy storage capacity of this space-filling carbon black network of the high specific surface area accessible to charge storage is shown to be an intensive quantity, whereas the high-rate capability of the carbon-cement electrodes exhibits self-similarity due to the hydration porosity available for charge transport. This intensive and self-similar nature of energy storage and rate capability represents an opportunity for mass scaling from electrode to structural scales. The availability, versatility, and scalability of these carbon-cement supercapacitors opens a horizon for the design of multifunctional structures that leverage high energy storage capacity, high-rate charge/discharge capabilities, and structural strength for sustainable residential and industrial applications ranging from energy autarkic shelters and self-charging roads for electric vehicles, to intermittent energy storage for wind turbines and tidal power stations.
ABSTRACT
Unlike the interface between two immiscible electrolyte solutions (ITIES) formed between water and polar solvents, molecular understanding of the liquid-liquid interface formed for aqueous biphasic systems (ABSs) is relatively limited and mostly relies on surface tension measurements and thermodynamic models. Here, high-resolution Raman imaging is used to provide spatial and chemical resolution of the interface of lithium chloride - lithium bis(trifluoromethanesulfonyl)imide - water (LiCl-LiTFSI-water) and HCl-LiTFSI-water, prototypical salt-salt ABSs found in a range of electrochemical applications. The concentration profiles of both TFSI anions and water are found to be sigmoidal thus not showing any signs of a positive adsorption for both salts and solvent. More striking, however, is the length at which the concentration profiles extend, ranging from 11 to 2 µm with increasing concentrations, compared to a few nanometers for ITIES. We thus reveal that unlike ITIES, salt-salt ABSs do not have a molecularly sharp interface but rather form an interphase with a gradual change of environment from one phase to the other. This knowledge represents a major stepping-stone in the understanding of aqueous interfaces, key for mastering ion or electron transfer dynamics in a wide range of biological and technological settings including novel battery technologies such as membraneless redox flow and dual-ion batteries.
ABSTRACT
Hydrogen hydrates are among the basic constituents of our solar system's outer planets, some of their moons, as well Neptune-like exo-planets. The details of their high-pressure phases and their thermodynamic conditions of formation and stability are fundamental information for establishing the presence of hydrogen hydrates in the interior of those celestial bodies, for example, against the presence of the pure components (water ice and molecular hydrogen). Here, we report a synthesis path and experimental observation, by X-ray diffraction and Raman spectroscopy measurements, of the most H[Formula: see text]-dense phase of hydrogen hydrate so far reported, namely the compound 3 (or C[Formula: see text]). The detailed characterisation of this hydrogen-filled ice, based on the crystal structure of cubic ice I (ice I[Formula: see text]), is performed by comparing the experimental observations with first-principles calculations based on density functional theory and the stochastic self-consistent harmonic approximation. We observe that the extreme (up to 90 GPa and likely beyond) pressure stability of this hydrate phase is due to the close-packed geometry of the hydrogen molecules caged in the ice I[Formula: see text] skeleton.
ABSTRACT
Current methods for detecting unlabeled antisense oligonucleotide (ASO) drugs rely on immunohistochemistry (IHC) and/or conjugated molecules, which lack sufficient sensitivity, specificity, and resolution to fully investigate their biodistribution. Our aim was to demonstrate the qualitative and quantitative distribution of unlabeled bepirovirsen, a clinical stage ASO, in livers and kidneys of dosed mice using novel staining and imaging technologies at subcellular resolution. ASOs were detected in formalin-fixed paraffin-embedded (FFPE) and frozen tissues using an automated chromogenic in situ hybridization (ISH) assay: miRNAscope. This was then combined with immunohistochemical detection of cell lineage markers. ASO distribution in hepatocytes versus nonparenchymal cell lineages was quantified using HALO AI image analysis. To complement this, hyperspectral coherent anti-Stokes Raman scattering (HS-CARS) imaging microscopy was used to specifically detect the unique cellular Raman spectral signatures following ASO treatment. Bepirovirsen was localized primarily in nonparenchymal liver cells and proximal renal tubules. Codetection of ASO with distinct cell lineage markers of liver and kidney populations aided target cell identity facilitating quantification. Positive liver signal was quantified using HALO AI, with 12.9% of the ASO localized to the hepatocytes and 87.1% in nonparenchymal cells. HS-CARS imaging specifically detected ASO fingerprints based on the unique vibrational signatures following unlabeled ASO treatment in a totally nonperturbative manner at subcellular resolution. Together, these novel detection and imaging modalities represent a significant increase in our ability to detect unlabeled ASOs in tissues, demonstrating improved levels of specificity and resolution. These methods help us understand their underlying mechanisms of action and ultimately improve the therapeutic potential of these important drugs for treating globally significant human diseases.
Subject(s)
Liver , Oligonucleotides, Antisense , Mice , Humans , Animals , Oligonucleotides, Antisense/genetics , Oligonucleotides, Antisense/metabolism , Tissue Distribution , Liver/diagnostic imaging , Liver/metabolism , In Situ Hybridization , Staining and LabelingABSTRACT
The time since deposition (TSD) of a bloodstain, i.e., the time of a bloodstain formation is an essential piece of biological evidence in crime scene investigation. The practical usage of some existing microscopic methods (e.g., spectroscopy or RNA analysis technology) is limited, as their performance strongly relies on high-end instrumentation and/or rigorous laboratory conditions. This paper presents a practically applicable deep learning-based method (i.e., BloodNet) for efficient, accurate, and costless TSD inference from a macroscopic view, i.e., by using easily accessible bloodstain photos. To this end, we established a benchmark database containing around 50,000 photos of bloodstains with varying TSDs. Capitalizing on such a large-scale database, BloodNet adopted attention mechanisms to learn from relatively high-resolution input images the localized fine-grained feature representations that were highly discriminative between different TSD periods. Also, the visual analysis of the learned deep networks based on the Smooth Grad-CAM tool demonstrated that our BloodNet can stably capture the unique local patterns of bloodstains with specific TSDs, suggesting the efficacy of the utilized attention mechanism in learning fine-grained representations for TSD inference. As a paired study for BloodNet, we further conducted a microscopic analysis using Raman spectroscopic data and a machine learning method based on Bayesian optimization. Although the experimental results show that such a new microscopic-level approach outperformed the state-of-the-art by a large margin, its inference accuracy is significantly lower than BloodNet, which further justifies the efficacy of deep learning techniques in the challenging task of bloodstain TSD inference. Our code is publically accessible via https://github.com/shenxiaochenn/BloodNet. Our datasets and pre-trained models can be freely accessed via https://figshare.com/articles/dataset/21291825.
Subject(s)
Blood Stains , Bayes Theorem , Machine LearningABSTRACT
Prenatal surgery for the treatment of spina bifida (myelomeningocele, MMC) significantly enhances the neurological prognosis of the patient. To ensure better protection of the spinal cord by large defects, the application of skin grafts produced with cells gained from the amniotic fluid is presently studied. In order to determine the most appropriate cells for this purpose, we tried to shed light on the extremely complex amniotic fluid cellular composition in healthy and MMC pregnancies. We exploited the potential of micro-Raman spectroscopy to analyse and characterize human amniotic fluid cells in total and putative (cKit/CD117-positive) stem cells of fetuses with MMC in comparison with amniotic fluid cells from healthy individuals, human fetal dermal fibroblasts and adult adipose derived stem cells. We found that (i) the differences between healthy and MMC amniocytes can be attributed to specific spectral regions involving collagen, lipids, sugars, tryptophan, aspartate, glutamate, and carotenoids, (ii) MMC amniotic fluid contains two particular cell populations which are absent or reduced in normal pregnancies, (iii) the cKit-negative healthy amniocyte subpopulation shares molecular features with human fetal fibroblasts. On the one hand we demonstrate a different amniotic fluid cellular composition in healthy and MMC pregnancies, on the other our work confirms micro-Raman spectroscopy to be a valuable tool for discriminating cell populations in unknown mixtures of cells.