A model of the scanning‐caused soft sample deformation during atomic force microscope contact‐mode (CM) imaging is developed. CM imaging has been widely used for topography characterization of live biological samples such as live cells. However, due to the intrinsic softness of these samples, the sample surface deformation caused by the probe–sample interaction leads to significant error in the topography images obtained, and damage to the live biological sample. Although the deformation can be reduced by imaging at a rather slow scan rate (e.g., less than 0.2 Hz), such a low‐speed imaging not only is time consuming, but also inevitably induces large temporal error. In this work, the scanning‐caused surface deformation of soft samples is modeled and quantified, including live biological samples in CM imaging. Effects of both the scanning and the coupling between the vertical and the lateral deformation on the deformation are characterized. The proposed deformation model is validated by implementing it to quantify the topography image difference caused by the scanning‐caused surface deformation of live prostate cancer cells imaged at two different (high and low) speeds, respectively.
Deep Learning for Live Cell Shape Detection and Automated AFM Navigation
Atomic force microscopy (AFM) provides a platform for high-resolution topographical imaging and the mechanical characterization of a wide range of samples, including live cells, proteins, and other biomolecules. AFM is also instrumental for measuring interaction forces and binding kinetics for protein–protein or receptor–ligand interactions on live cells at a single-molecule level. However, performing force measurements and high-resolution imaging with AFM and data analytics are time-consuming and require special skill sets and continuous human supervision. Recently, researchers have explored the applications of artificial intelligence (AI) and deep learning (DL) in the bioimaging field. However, the applications of AI to AFM operations for live-cell characterization are little-known. In this work, we implemented a DL framework to perform automatic sample selection based on the cell shape for AFM probe navigation during AFM biomechanical mapping. We also established a closed-loop scanner trajectory control for measuring multiple cell samples at high speed for automated navigation. With this, we achieved a 60× speed-up in AFM navigation and reduced the time involved in searching for the particular cell shape in a large sample. Our innovation directly applies to many bio-AFM applications with AI-guided intelligent automation through image data analysis together with smart navigation.
more »
« less
- Award ID(s):
- 1751503
- NSF-PAR ID:
- 10428261
- Date Published:
- Journal Name:
- Bioengineering
- Volume:
- 9
- Issue:
- 10
- ISSN:
- 2306-5354
- Page Range / eLocation ID:
- 522
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
more » « less
Abstract -
more » « less
Modern high-throughput nanopatterning techniques, such as nanoimprint lithography, make it possible to fabricate arrays of nanostructures (features with dimensions of 10’s to 100’s of nm) over large area substrates (cm2to m2scale) such as Si wafers, glass sheets, and flexible roll-to-roll webs. The ability to make such large-area nanostructure arrays (LNAs) has created an extensive design space, enabling a wide array of applications including optical devices, such as wire-grid polarizers, transparent conductors, color filters, and anti-reflection surfaces, and building blocks for electronic components, such as ultracapacitors, sensors, and memory storage architectures. However, existing metrology methods will have trouble scaling alongside fabrication methods. Scanning electron microscopy (SEM) and atomic force microscopy (AFM), for instance, have micron scale fields of view (FOV) that preclude comprehensive characterization of LNAs, which may be manufactured at m2per minute rates. Scatterometry approaches have larger FOVs (typically 100’s of µm to a few mm), but traditional scatterometry systems measure samples one point at a time, which also makes them too slow for large-scale LNA manufacturing. In this work, we demonstrate parallelization of the traditional spectroscopic scatterometry approach using hyperspectral imaging, increasing the throughput of the technique by a factor of 106-107. We demonstrate this approach by using hyperspectral imaging and inverse modeling of reflectance spectra to derive 3-dimensional geometric data for Si nanopillar array structures over both mm and cm-scale with µm-scale spatial resolution. This work suggests that geometric measurements for a variety of LNAs can be performed with the potential for high speed over large areas which may be critical for future LNA manufacturing.
-
Atomic Force Microscopy (AFM) is widely used for topographic imaging of DNA and protein-DNA complexes in ambient conditions with nanometer resolution. In AFM studies of protein-DNA complexes, identifying the protein’s location on the DNA substrate is one of the major goals. Such studies require distinguishing between the DNA ends, which can be accomplished by end-specific labeling of the DNA substrate. We selected as labels three-way DNA junctions (3WJ) assembled from synthetic DNA oligonucleotides with two arms of 39–40 bp each. The third arm has a three-nucleotide overhang, GCT, which is paired with the sticky end of the DNA substrate generated by the SapI enzyme. Ligation of the 3WJ results in the formation of a Y-type structure at the end of the linear DNA mole cule, which is routinely identified in the AFM images. The yield of labeling is 69%. The relative orientation of arms in the Y-end varies, such dynamics were directly visualized with time-lapse AFM studies using high-speed AFM (HS-AFM). This labeling approach was applied to the characterization of the nucleosome arrays assembled on different DNA templates. HS-AFM experiments revealed a high dynamic of nucleosomes resulting in a spontaneous unraveling followed by disassembly of nucleosomes.more » « less
-
Fadlelmola, Faisal Mohamed (Ed.)Sickle cell disease, a genetic disorder affecting a sizeable global demographic, manifests in sickle red blood cells (sRBCs) with altered shape and biomechanics. sRBCs show heightened adhesive interactions with inflamed endothelium, triggering painful vascular occlusion events. Numerous studies employ microfluidic-assay-based monitoring tools to quantify characteristics of adhered sRBCs from high resolution channel images. The current image analysis workflow relies on detailed morphological characterization and cell counting by a specially trained worker. This is time and labor intensive, and prone to user bias artifacts. Here we establish a morphology based classification scheme to identify two naturally arising sRBC subpopulations—deformable and non-deformable sRBCs—utilizing novel visual markers that link to underlying cell biomechanical properties and hold promise for clinically relevant insights. We then set up a standardized, reproducible, and fully automated image analysis workflow designed to carry out this classification. This relies on a two part deep neural network architecture that works in tandem for segmentation of channel images and classification of adhered cells into subtypes. Network training utilized an extensive data set of images generated by the SCD BioChip, a microfluidic assay which injects clinical whole blood samples into protein-functionalized microchannels, mimicking physiological conditions in the microvasculature. Here we carried out the assay with the sub-endothelial protein laminin. The machine learning approach segmented the resulting channel images with 99.1±0.3% mean IoU on the validation set across 5 k -folds, classified detected sRBCs with 96.0±0.3% mean accuracy on the validation set across 5 k -folds, and matched trained personnel in overall characterization of whole channel images with R 2 = 0.992, 0.987 and 0.834 for total, deformable and non-deformable sRBC counts respectively. Average analysis time per channel image was also improved by two orders of magnitude (∼ 2 minutes vs ∼ 2-3 hours) over manual characterization. Finally, the network results show an order of magnitude less variance in counts on repeat trials than humans. This kind of standardization is a prerequisite for the viability of any diagnostic technology, making our system suitable for affordable and high throughput disease monitoring.more » « less
-
Jabbari, Esmaiel (Ed.)This study presents novel biocompatible Polydimethylsiloxane (PDMS)-based micromechanical tweezers (μTweezers) capable of the stiffness characterization and manipulation of hydrogel-based organoids. The system showed great potential for complementing established mechanical characterization methods such as Atomic Force Microscopy (AFM), parallel plate compression (PPC), and nanoindentation, while significantly reducing the volume of valuable hydrogels used for testing. We achieved a volume reduction of ~0.22 μl/sample using the μTweezers vs. ~157 μl/sample using the PPC, while targeting high-throughput measurement of widely adopted micro-mesoscale (a few hundred μm-1500 μm) 3D cell cultures. The μTweezers applied and measured nano-millinewton forces through cantilever’ deflection with high linearity and tunability for different applications; the assembly is compatible with typical inverted optical microscopes and fit on standard tissue culture Petri dishes, allowing mechanical compression characterization of arrayed 3D hydrogel-based organoids in a high throughput manner. The average achievable output per group was 40 tests per hour, where 20 organoids and 20 reference images in one 35 mm petri dish were tested, illustrating efficient productivity to match the increasing demand on 3D organoids’ applications. The changes in stiffness of collagen I hydrogel organoids in four conditions were measured, with ovarian cancer cells (SKOV3) or without (control). The Young’s modulus of the control group (Control—day 0, E = 407± 146, n = 4) measured by PPC was used as a reference modulus, where the relative elastic compressive modulus of the other groups based on the stiffness measurements was also calculated (control-day 0, E = 407 Pa), (SKOV3-day 0, E = 318 Pa), (control-day 5, E = 528 Pa), and (SKOV3-day 5, E = 376 Pa). The SKOV3-embedded hydrogel-based organoids had more shrinkage and lowered moduli on day 0 and day 5 than controls, consistently, while SKOV3 embedded organoids increased in stiffness in a similar trend to the collagen I control from day 0 to day 5. The proposed method can contribute to the biomedical, biochemical, and regenerative engineering fields, where bulk mechanical characterization is of interest. The μTweezers will also provide attractive design and application concepts to soft membrane-micro 3D robotics, sensors, and actuators.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.3390/bioengineering9100522
