fluorescence image preprocessing  (MathWorks Inc)

Journal: Biomedical Optics Express

Article Title: Temporal variance mapping with machine learning for label-free 3D chromatin imaging using optical interferometric microscopy

doi: 10.1364/BOE.583584

Figure Lengend Snippet: Machine learning-based inference of chromatin fluorescence from label-free structure and dynamics maps. (a) Schematic of the training and validation process for a CNN model f S D from paired label-free inputs ( S , D ) to fluorescence outputs f S D ( S , D ) . (b) Quantitative evaluation of prediction accuracy using MS - SSIM ( Y ˆ , Y ) to compare the model-generated fluorescence images Y ˆ , with ground truth confocal images Y . In the box plots, the lower and upper boundaries of the box indicate the 25th and 75th percentiles, respectively. Within the box, a solid line marks the median. [*** P < 0.001 (Student’s t test)] (c–f) Confocal fluorescence image of H2B-mCherry as ground truth Y (c), and CNN predictions obtained with combined inputs Y ˆ S D = f S D ( S , D ) , (d), dynamics map only Y ˆ D = f D ( D ) , (e), and structure map only Y ˆ S = f S ( S ) , (f) as input.

Article Snippet: Fluorescence image preprocessing was performed in MATLAB using a custom-built pipeline designed to reduce noise, normalize intensity, and register images for in silico labeling (see Fig. S3 for the flowchart of image processing).

Techniques: Fluorescence, Biomarker Discovery, Generated

Journal: Biomedical Optics Express

Article Title: Temporal variance mapping with machine learning for label-free 3D chromatin imaging using optical interferometric microscopy

doi: 10.1364/BOE.583584

Figure Lengend Snippet: Label-free, high-resolution imaging of nuclear structure and dynamics in live cells. (a) Schematic of the COBRI microscope integrated with a confocal fluorescence channel, enabling simultaneous label-free and fluorescence imaging. (b) COBRI image of a live cell nucleus showing clearly resolved nucleoli (indicated by arrows). (c) Temporal variance map of the same nucleus, calculated from a sequence of COBRI images acquired at 1,000 frames per second. The spatial variation in variance reflects differences in scattering intensity caused by dynamic molecular structures. (d) Confocal fluorescence image of the same nucleus expressing H2B-mCherry, illustrating the spatial distribution of chromatin for reference.

Article Snippet: Fluorescence image preprocessing was performed in MATLAB using a custom-built pipeline designed to reduce noise, normalize intensity, and register images for in silico labeling (see Fig. S3 for the flowchart of image processing).

Techniques: Imaging, Microscopy, Fluorescence, Sequencing, Expressing

Journal: Biomedical Optics Express

Article Title: Temporal variance mapping with machine learning for label-free 3D chromatin imaging using optical interferometric microscopy

doi: 10.1364/BOE.583584

Figure Lengend Snippet: Correlation between dynamics maps and confocal fluorescence images of chromatin. (a) Joint intensity distribution of the pixel-wise differential temporal variance values from the dynamics map (computed from 2,000 frames) and fluorescence intensities of H2B-mCherry, showing a modest positive correlation. (b) Correlation between the dynamics maps computed with varying numbers of frames and the reference dynamics map computed with 2,000 frames. All image data are recorded at 1000 Hz with a frame time of 1 ms. (c) Representative dynamics maps of a cell nucleus computed using increasing numbers of frames, ranging from 20 to 2,000. (d) Confocal fluorescence image of H2B-mCherry from the same nucleus shown in (c).

Article Snippet: Fluorescence image preprocessing was performed in MATLAB using a custom-built pipeline designed to reduce noise, normalize intensity, and register images for in silico labeling (see Fig. S3 for the flowchart of image processing).

Techniques: Fluorescence

Journal: Biomedical Optics Express

Article Title: Temporal variance mapping with machine learning for label-free 3D chromatin imaging using optical interferometric microscopy

doi: 10.1364/BOE.583584

Figure Lengend Snippet: Enhanced axial resolution in dynamics maps. (a) Confocal fluorescence image (left), structure map (middle), and dynamics map (right) of a cell nucleus acquired at two focal planes separated by 1.2 µm. Arrows indicate the lateral position of a nucleolus, which is in focus at Δz = 0. When the focal plane is shifted 1.2 µm above, the nucleolus becomes undetectable in both the confocal fluorescence and dynamics maps but remains visible in the structure map. (b) xz cross-sections of the nucleus along the line indicated in (a). The out-of-focus shadow in the structure map, highlighted by the asterisk symbols, underscores its limited axial resolution. In contrast, the dynamics map demonstrates markedly improved axial resolution.

Article Snippet: Fluorescence image preprocessing was performed in MATLAB using a custom-built pipeline designed to reduce noise, normalize intensity, and register images for in silico labeling (see Fig. S3 for the flowchart of image processing).

Techniques: Fluorescence

Journal: Biomedical Optics Express

Article Title: Temporal variance mapping with machine learning for label-free 3D chromatin imaging using optical interferometric microscopy

doi: 10.1364/BOE.583584

Figure Lengend Snippet: Visualization of nuclear speckles using the dynamics map. (a) Structure map of a cell nucleus, with nucleoli clearly visible and outlined by white dashed circles. (b) Corresponding dynamics map of the same nucleus, where both nucleoli (white dashed circles) and nuclear speckles (blue arrows) are identified based on their distinct fluctuation patterns. (c) Confocal fluorescence image of SRRM2-mCherry, highlighting nuclear speckles as bright puncta (blue arrows). The nuclear boundary is marked with a green dashed line in all panels.

Article Snippet: Fluorescence image preprocessing was performed in MATLAB using a custom-built pipeline designed to reduce noise, normalize intensity, and register images for in silico labeling (see Fig. S3 for the flowchart of image processing).

Techniques: Fluorescence

Journal: Biomedical Optics Express

Article Title: Temporal variance mapping with machine learning for label-free 3D chromatin imaging using optical interferometric microscopy

doi: 10.1364/BOE.583584

Figure Lengend Snippet: Predicted confocal fluorescence images of chromatin and their corresponding ground truths. (a) Seven representative paired images of confocal fluorescence signals predicted by our algorithm using structural and dynamical maps as input, alongside their ground truth counterparts. (b) Joint intensity distribution of pixel-wise predicted fluorescence values and H2B-mCherry fluorescence intensities (outliers removed), demonstrating a strong positive correlation with improved linearity.

Article Snippet: Fluorescence image preprocessing was performed in MATLAB using a custom-built pipeline designed to reduce noise, normalize intensity, and register images for in silico labeling (see Fig. S3 for the flowchart of image processing).

Techniques: Fluorescence