- Award ID(s):
- 2000554
- NSF-PAR ID:
- 10316487
- Date Published:
- Journal Name:
- Frontiers in Computer Science
- Volume:
- 3
- ISSN:
- 2624-9898
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
In the field of materials science, microscopy is the first and often only accessible method for structural characterization. There is a growing interest in the development of machine learning methods that can automate the analysis and interpretation of microscopy images. Typically training of machine learning models requires large numbers of images with associated structural labels, however, manual labeling of images requires domain knowledge and is prone to human error and subjectivity. To overcome these limitations, we present a semi-supervised transfer learning approach that uses a small number of labeled microscopy images for training and performs as effectively as methods trained on significantly larger image datasets. Specifically, we train an image encoder with unlabeled images using self-supervised learning methods and use that encoder for transfer learning of different downstream image tasks (classification and segmentation) with a minimal number of labeled images for training. We test the transfer learning ability of two self-supervised learning methods: SimCLR and Barlow-Twins on transmission electron microscopy (TEM) images. We demonstrate in detail how this machine learning workflow applied to TEM images of protein nanowires enables automated classification of nanowire morphologies ( e.g. , single nanowires, nanowire bundles, phase separated) as well as segmentation tasks that can serve as groundwork for quantification of nanowire domain sizes and shape analysis. We also extend the application of the machine learning workflow to classification of nanoparticle morphologies and identification of different type of viruses from TEM images.more » « less
-
Various biomacromolecule components of extracellular matrix (ECM) link together to form a structurally stable composite. Monitoring of such matrix microstructure can be very important in studying structure-associated cellular processes, improving cellular function, and ensuring sufficient mechanical integrity in engineered tissues. This paper describes a novel method to study microscale alignment of matrix in engineered tissue scaffolds (ETS) that were usually composed of a variety of biomacromolecules derived by cells. as the organization of overall biomacromolecule network has been seldomly examined. First, a trained loading function was derived from Raman spectra of highly aligned native tissue via PCA, where prominent changes associated with Raman bands (e.g., 1444, 1465, 1605, 1627-1660 and 1665-1689 cm−1) were detected with respect to the polarized angle. These changes were mainly caused by the aligned matrix of many compounds within the tissue relative to the laser polarization, including proteins, lipids and carbohydrates. Hence this trained function was applied to quantify the alignment within ETS of various matrix components derived by cells. A simple metric called Amplitude Alignment Metric was derived to correlate the orientation dependence of polarized Raman spectra of ETS to the degree of matrix alignment. By acquiring polarized Raman spectra of ETS at micrometer regions, the Amplitude Alignment Metric was significantly higher in anisotropic ETS than isotropic ones. The PRS method showed a lower p-value for distinguishing the alignment between the two types of ETS as compared to the microscopic method for detecting fluorescently labeled protein matrices at similar microscopic scale. These results indicate the anisotropy of complex matrix in engineered tissue can be assessed at microscopic scale using a PRS-based simple metric, superior to traditional microscopic method. This PRS-based method can serve as a complementary tool for the design and assessment of engineered tissues that mimic the native matrix organizational microstructures.more » « less
-
Cancer cell migration is significantly influenced by the physical properties (e.g., mechanics, topography) of the local extracellular matrix (ECM). However, how cells guide their movement in a complex, heterogeneous landscape while interpreting various physical factors, especially those with sizes on the subcellular length scale, is not completely understood. Herein, a photodegradable hydrogel is fabricated with control over the spatial distribution of features with different physical characteristics. On the surface of this hydrogel, patterns of features with sizes and spacings on the subcellular to cellular length scale are generated using a digital micromirror device, and the influence of the organization of these patterns on cellular migratory behavior is investigated. The size and spatial arrangement of hard (or stiff) and soft domains, more so than their topography, is found to have a significant impact on the migration speed. Using this approach, the maximum migration speed is observed on hydrogels with 10 μm pattern sizes spaced 10 μm apart. In this regime, the cells are confined and interact with both hard and soft regions at their periphery. Together, this study of cell–ECM interactions at the interface of hard and soft domains provides insights into how cancer cells interact with their environment.
-
Abstract Tuning cell shape by altering the biophysical properties of biomaterial substrates on which cells operate would provide a potential shape-driven pathway to control cell phenotype. However, there is an unexplored dimensional scale window of three-dimensional (3D) substrates with precisely tunable porous microarchitectures and geometrical feature sizes at the cell’s operating length scales (10–100 μm). This paper demonstrates the fabrication of such high-fidelity fibrous substrates using a melt electrowriting (MEW) technique. This advanced manufacturing approach is biologically qualified with a metrology framework that models and classifies cell confinement states under various substrate dimensionalities and architectures. Using fibroblasts as a model cell system, the mechanosensing response of adherent cells is investigated as a function of variable substrate dimensionality (2D vs. 3D) and porous microarchitecture (randomly oriented, “non-woven” vs. precision-stacked, “woven”). Single-cell confinement states are modeled using confocal fluorescence microscopy in conjunction with an automated single-cell bioimage data analysis workflow that extracts quantitative metrics of the whole cell and sub-cellular focal adhesion protein features measured. The extracted multidimensional dataset is employed to train a machine learning algorithm to classify cell shape phenotypes. The results show that cells assume distinct confinement states that are enforced by the prescribed substrate dimensionalities and porous microarchitectures with the woven MEW substrates promoting the highest cell shape homogeneity compared to non-woven fibrous substrates. The technology platform established here constitutes a significant step towards the development of integrated additive manufacturing—metrology platforms for a wide range of applications including fundamental mechanobiology studies and 3D bioprinting of tissue constructs to yield specific biological designs qualified at the single-cell level.
-
ABSTRACT BACKGROUND AND PURPOSE Neurosurgical resection is one of the few opportunities researchers have to image the human brain pre‐ and postfocal damage. A major challenge associated with brains undergoing surgical resection is that they often do not fit brain templates most image‐processing methodologies are based on. Manual intervention is required to reconcile the pathology, requiring time investment and introducing reproducibility concerns, and extreme cases must be excluded.
METHODS We propose an automatic longitudinal pipeline based on High Angular Resolution Diffusion Imaging acquisitions to facilitate a Pathway Lesion Symptom Mapping analysis relating focal white matter injury to functional deficits. This two‐part approach includes (i) automatic segmentation of focal white matter injury from anisotropic power differences, and (ii) modeling disconnection using tractography on the single‐subject level, which specifically identifies the disconnections associated with focal white matter damage.
RESULTS The advantages of this approach stem from (1) objective and automatic lesion segmentation and tractogram generation, (2) objective and precise segmentation of affected tissue likely to be associated with damage to long‐range white matter pathways (defined by anisotropic power), (3) good performance even in the cases of anatomical distortions by use of nonlinear tensor‐based registration, which aligns images using an approach sensitive to white matter microstructure.
CONCLUSIONS Mapping a system as variable and complex as the human brain requires sample sizes much larger than the current technology can support. This pipeline can be used to execute large‐scale, sufficiently powered analyses by meeting the need for an automatic approach to objectively quantify white matter disconnection.