Dataset Information

Deep learning-based point-scanning super-resolution imaging.

ABSTRACT: Point-scanning imaging systems are among the most widely used tools for high-resolution cellular and tissue imaging, benefiting from arbitrarily defined pixel sizes. The resolution, speed, sample preservation and signal-to-noise ratio (SNR) of point-scanning systems are difficult to optimize simultaneously. We show these limitations can be mitigated via the use of deep learning-based supersampling of undersampled images acquired on a point-scanning system, which we term point-scanning super-resolution (PSSR) imaging. We designed a 'crappifier' that computationally degrades high SNR, high-pixel resolution ground truth images to simulate low SNR, low-resolution counterparts for training PSSR models that can restore real-world undersampled images. For high spatiotemporal resolution fluorescence time-lapse data, we developed a 'multi-frame' PSSR approach that uses information in adjacent frames to improve model predictions. PSSR facilitates point-scanning image acquisition with otherwise unattainable resolution, speed and sensitivity. All the training data, models and code for PSSR are publicly available at 3DEM.org.

SUBMITTER: Fang L

PROVIDER: S-EPMC8035334 | biostudies-literature | 2021 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deep learning-based point-scanning super-resolution imaging.

Fang Linjing L Monroe Fred F Novak Sammy Weiser SW Kirk Lyndsey L Schiavon Cara R CR Yu Seungyoon B SB Zhang Tong T Wu Melissa M Kastner Kyle K Latif Alaa Abdel AA Lin Zijun Z Shaw Andrew A Kubota Yoshiyuki Y Mendenhall John J Zhang Zhao Z Pekkurnaz Gulcin G Harris Kristen K Howard Jeremy J Manor Uri U

Nature methods 20210308 4

Point-scanning imaging systems are among the most widely used tools for high-resolution cellular and tissue imaging, benefiting from arbitrarily defined pixel sizes. The resolution, speed, sample preservation and signal-to-noise ratio (SNR) of point-scanning systems are difficult to optimize simultaneously. We show these limitations can be mitigated via the use of deep learning-based supersampling of undersampled images acquired on a point-scanning system, which we term point-scanning super-reso ...[more]

PMID: 33686300

Similar Datasets

Project description:PurposeTo develop a super-resolution technique using convolutional neural networks for generating thin-slice knee MR images from thicker input slices, and compare this method with alternative through-plane interpolation methods.MethodsWe implemented a 3D convolutional neural network entitled DeepResolve to learn residual-based transformations between high-resolution thin-slice images and lower-resolution thick-slice images at the same center locations. DeepResolve was trained using 124 double echo in steady-state (DESS) data sets with 0.7-mm slice thickness and tested on 17 patients. Ground-truth images were compared with DeepResolve, clinically used tricubic interpolation, and Fourier interpolation methods, along with state-of-the-art single-image sparse-coding super-resolution. Comparisons were performed using structural similarity, peak SNR, and RMS error image quality metrics for a multitude of thin-slice downsampling factors. Two musculoskeletal radiologists ranked the 3 data sets and reviewed the diagnostic quality of the DeepResolve, tricubic interpolation, and ground-truth images for sharpness, contrast, artifacts, SNR, and overall diagnostic quality. Mann-Whitney U tests evaluated differences among the quantitative image metrics, reader scores, and rankings. Cohen's Kappa (κ) evaluated interreader reliability.ResultsDeepResolve had significantly better structural similarity, peak SNR, and RMS error than tricubic interpolation, Fourier interpolation, and sparse-coding super-resolution for all downsampling factors (p < .05, except 4 × and 8 × sparse-coding super-resolution downsampling factors). In the reader study, DeepResolve significantly outperformed (p < .01) tricubic interpolation in all image quality categories and overall image ranking. Both readers had substantial scoring agreement (κ = 0.73).ConclusionDeepResolve was capable of resolving high-resolution thin-slice knee MRI from lower-resolution thicker slices, achieving superior quantitative and qualitative diagnostic performance to both conventionally used and state-of-the-art methods.

Project description:Over the last decade, single-molecule localization microscopy (SMLM) has developed into a set of powerful techniques that have improved spatial resolution over diffraction-limited microscopy and demonstrated the ability to resolve biological features down to a few tens of nanometers. We introduce a single molecule-based scanning SMLM (scanSMLM) system that enables rapid volume imaging. Along with epi-illumination, the system employs a scanning-based 4f detection for volume imaging. The 4f system comprises a combination of an electrically-tunable lens and high NA detection objective lens. By rapidly changing the aperture (or equivalently the focus) of an electrically-tunable lens (ETL) in a 4f detection system, the selectivity of the axial object plane is achieved, for which the image forms in the image/detector plane. So, in principle, one can scan the object volume by just altering the aperture of ETL. Two schemes were adopted to carry out volume imaging: cyclic scan and conventional scan. The cyclic scheme scans the volume in each scan cycle, whereas plane-wise scanning is performed in the conventional scheme. Hence, the cyclic scan ensures uniform dwell time on each frame during data collection, thereby evenly distributing photobleaching throughout the cell volume. With a minimal change in the system hardware (requiring the addition of an ETL lens and related electronics for step-voltage generation) in the existing SMLM system, volume scanning (along the z-axis) can be achieved. To calibrate and derive critical system parameters, we imaged fluorescent beads embedded in a gel-matrix 3D block as a test sample. Subsequently, scanSMLM is employed to visualize the architecture of actin-filaments and the distribution of Meos-Tom20 molecules on the mitochondrial membrane. The technique is further exploited to understand the clustering of Hemagglutinin (HA) protein single molecules in a transfected cell for studying Influenza-A disease progression. The system, for the first time, enabled 3D visualization of HA distribution that revealed HA cluster formation spanning the entire cell volume, post 24 hrs of transfection. Critical biophysical parameters related to HA clusters (density, the number of HA molecules per cluster, axial span, fraction of clustered molecules, and others) are also determined, giving an unprecedented insight into Influenza-A disease progression at the single-molecule level.

Dataset Information

Deep learning-based point-scanning super-resolution imaging.

Publications

Deep learning-based point-scanning super-resolution imaging.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets