Journal of microscopy最新文献

Transition metal dichalcogenides (TMDs) are promising two-dimensional materials whose properties are strongly influenced by substrate interactions. While conventional Raman spectroscopy probes these effects, its diffraction-limited resolution often averages out local variations such as strain, masking intrinsic behaviours. Here, we employ tip-enhanced Raman spectroscopy (TERS) to investigate the vibrational properties of monolayer $�{\mathrm{WS}}_2$ and $�{\mathrm{MoSe}}_2$ on top of glass and glass/hBN substrates. TERS offers nanometric spatial resolution, allowing direct correlation between Raman features and topographical inhomogeneities. Our results reveal that local variations in strain, doping, and dielectric screening that vary across the substrate interface are often accompanied by nanoscale structural features such as wrinkles, which locally modulate the vibrational response.

Pattern recognition in convolutional neural networks (CNNs) is computationally intensive due to its reliance on 2D convolutions, requiring significant processing power and time. This paper proposes an analogue optical hardware system to improve CNN efficiency, focusing on forward propagation tasks such as data preparation, correlation, and decision-making. By utilising the continuous properties of light waves for 2D convolutional operations, the system overcomes key limitations of von Neumann architectures around saving power and time. Optical wave operations allow for more efficient and instantaneous tasks like 2D Fourier transforms, which are crucial to pattern recognition. The paper validates these concepts through simulations using MATLAB and COMSOL. Overall, the presented approach paves the way for more efficient ML hardware. Future work will focus on extending the system to enable full CNN training, including backward propagation, as well as the development of commercially suitable hardware implementations.

In situ synchrotron X-ray computed tomography enables dynamic material studies. However, automated segmentation remains challenging due to complex imaging artefacts - like ring and cupping effects - and limited training data. We present a methodology for deep learning-based segmentation by transforming high-quality ex situ laboratory data to train models for segmentation of in situ synchrotron data, demonstrated through a metal oxide dissolution study. Using a modified SegFormer architecture, our approach achieves segmentation performance (94.7% IoU) that matches human inter-annotator reliability (94.6% IoU). This indicates the model has reached the practical upper bound for this task, while reducing processing time by 2 orders of magnitude per 3D dataset compared to manual segmentation. The method maintains robust performance over significant morphological changes during experiments, despite training only on static specimens. This methodology can be readily applied to diverse materials systems, enabling the efficient analysis of the large volumes of time-resolved tomographic data generated in typical in situ experiments across scientific disciplines.

The comprehensive characterisation of complex, irreplaceable cultural heritage artefacts presents significant challenges for traditional analytical methods, which can fall short in providing multi-scale, non-invasive analysis. Correlative Multimodal Microscopy (CoMic), an approach that integrates data from multiple techniques, offers a powerful solution by bridging structural, chemical, and topographical information across different length scales. This paper provides a comprehensive review of the evolution, current applications, and future trajectory of CoMic within the field of heritage science. We present a historical overview of microscopy in heritage studies and detail the principles and advances of key techniques, such as electron, X-ray, optical, and probe microscopies. This review presents practical applications through case studies on materials that include wood, pigments, ceramics, metals, and textiles. To aid CoMic uptake, we also provide user-centric guides for researchers with diverse expertise. This review also examines the challenges that currently limit the widespread adoption of CoMic, challenges that include sample preparation, data correlation accuracy, high instrumental and resource costs, and the need for specialised interdisciplinary expertise. Although CoMic is a transformative methodology for artefact analysis and conservation, its full potential will be realised through future developments in accessible instrumentation, standardised protocols, and the integration of AI-driven data analysis. This review serves as a critical resource and roadmap for researchers, conservators, and institutions looking to harness the power of correlative microscopy to preserve our shared cultural legacy.

Collagen, a key structural component of the extracellular matrix, assembles through a hierarchical process of fibrillogenesis. Despite extensive studies on mature collagen fibrils, intermediates such as protofibrils remain underexplored, particularly at the nanoscale. This study presents hyperspectral tip-enhanced Raman spectroscopy (TERS) imaging of collagen protofibrils, offering chemical and structural insights into early fibrillogenesis by acquiring nanoscale molecular profiles of collagen intermediates. TERS spectra, complemented by atomic force microscopy (AFM) images, reveal characteristic molecular vibrational modes, including the phenylalanine ring breathing mode, amide II and $�{\mathrm{CH}}_2$/$�{\mathrm{CH}}_3$ stretching vibrations, with distinct spectral signatures compared to mature fibrils.

Bioimaging has transformed our understanding of biological processes, yet extracting meaningful information from complex datasets remains a challenge, particularly for biologists without computational expertise. This paper proposes a simple general approach, to help identify which image analysis methods could be relevant for a given image dataset. We first categorise structures commonly observed in bioimage data into different types related to image analysis domains. Based on these types, we provide a list of methods adapted to the quantification of images from each category. Our approach includes illustrative examples and a visual flowchart, to help researchers define analysis objectives clearly. By understanding the diversity of bioimage structures and linking them with appropriate analysis approaches, the framework empowers researchers to navigate bioimage datasets more efficiently. It also aims to foster a common language between researchers and analysts, thereby enhancing mutual understanding and facilitating effective communication.

The contributions of coherent bright-field phase and incoherent dark-field amplitude contrast are investigated for thick biological specimens. A model for a T4 phage is constructed and images simulated for both TEM and STEM phase contrast using a multislice code. For TEM, the fraction of the illumination intensity available for phase contrast imaging is limited by the fraction of electrons in the zero loss peak, the plasmon peak, or the Landau distribution peak for very thick specimens. These were measured from electron energy loss spectra recorded from various thicknesses of vitreous ice. The incoherent amplitude contrast is simulated using the Penelope Monte Carlo code. Noise limits the features that can be distinguished under the low-dose conditions required for cryo-EM, even for high electron exposures of 100 electrons/Ų. Since in STEM post specimen optics are not used to form the image inelastically scattered electrons contribute to the recorded intensity. In principle STEM should have an advantage over TEM not just for incoherent amplitude contrast but also for coherent phase contrast beyond the limit of weak phase. The simulations suggest that it should be possible to image features in the phage embedded in 1 µm of vitreous ice when collection angles are optimised for bright or dark-field signals, with best contrast achieved for accelerating voltages of about 700 keV.

Anomalously low values of the normalised variance in fluctuation electron microscopy (FEM) have been frequently reported. We present three experimental corrections for quantitative interpretation that significantly modify conventional approaches. FEM relies on measurements of intensity statistics in coherent nanodiffraction patterns. We demonstrate that sampling the nanodiffraction patterns with a pixelated detector removes high-frequency signals and reduces statistical variance. The most significant impact is on the background normalised variance, which arises from random atomic alignments and is distinct from the normalised variance peaks associated with the correlated alignments of medium-range order. Indeed, we show that if the peaks are background-subtracted, their height is much less affected by the detector effect, provided the experimental conditions are optimised. We show that shot noise correction must also be adjusted to account for the camera Modulation Transfer Function (MTF) effects. Additionally, we demonstrate through experiment that the traditional method of thickness correction for a-Si is inadequate and propose an alternative approach to address thickness variations. We speculate on the origin of the anomalous thickness effect in terms of displacement decoherence due to sample ‘fluttering’ under irradiation.

The pore structure characteristics of coal are crucial for coalbed methane adsorption and migration, carbon storage, and safety in deep coal mining. Although traditional methods can detect pore volume and distribution, they are limited in analysing pore morphology and surface properties. This study employs multiscale techniques including AFM (Atomic force microscopy), SEM (Scanning electron microscopy), and LP-N₂GA (Low-Pressure nitrogen gas adsorption) to systematically analyse the impact of coal rank changes on pore structure and its evolutionary process, covering coals from medium-volatile to low-volatile bituminous and anthracite coals. AFM reveals the three-dimensional morphology and quantitative parameters of nanopores, SEM observes meso- and micropore structures, and LP-N₂GA verifies pore size distribution. As coal rank increases, surface roughness decreases significantly, the number of pores increases, the average pore diameter decreases, pore morphology transforms from irregular to circular, and porosity increases. Specifically, as the rank of coal increases, the number of nanoring structures rises, while their diameters decrease. Changes in coal rank profoundly affect the nanoring structure, consistent with the evolutionary trend of surface morphology. The combination of AFM and LP-N₂GA reveals the role of micropores in gas adsorption. This research not only provides a new perspective for understanding the influence of coal rank changes on pore structure characteristics but also offers a theoretical foundation for coalbed methane development, geological sequestration of carbon dioxide, design of coal-based functional materials, and coal mine safety prevention and control.

Structured illumination microscopy (SIM) as a type of super-resolution optical microscopy technique has been widely used in the fields of biophysics, neuroscience, and cell biology research. However, this technique often requires high-intensity illumination and multiple image acquisitions to generate a single high-resolution image. This process not only significantly reduces the imaging speed, but also increases the exposure time of samples to intense light, leading to increased phototoxicity and photobleaching issues, especially prominent in live cell imaging. Here, we propose a lightweight Multi-Convolutional UNet (MCU-Net) aiming to maintain efficient super-resolution reconstruction performance by reducing the model parameter quantity. The algorithm integrates multiple convolutional techniques with multi-scale attention mechanisms, enhancing the model's sensitivity to information at different scales and improving its precise recognition ability for image textures and structures, thus enabling high-quality super-resolution reconstruction even under low-light conditions. The overall performance of the model is evaluated in terms of efficiency and accuracy, comparing MCU-Net with deep neural network models (UNet, ScUNet, EDSR, DFCAN) and traditional reconstruction algorithms (Wiener, HiFi, TV) across different cell types, lighting intensities, and various test sets. Experimental results show that compared to other deep learning models, MCU-Net achieves a 12.66% improvement in MS-SSIM and a 50.79% increase in NRMSE index. Its prediction accuracy remains stable even in the presence of low signal-to-noise ratio inputs. Furthermore, it strikes an optimal balance between reconstruction speed and model accuracy, with a 76.10% improvement in inference speed compared to the DFCAN model.