Microsnoop: A Generalized Tool for Unbiased Representation of Diverse Microscopy Images

Abstract

Accurate and automated representation of microscopy images from small-scale to high-throughput is becoming an essential procedure in basic and applied biological research. Here, we present Microsnoop, a novel deep learning-based representation tool trained on large-scale microscopy images using masked self-supervised learning, which eliminates the need for manual annotation. Microsnoop is able to unbiasedly profile a wide range of complex and heterogeneous images, including single-cell, fully-imaged and batch-experiment data. We evaluated the performance of Microsnoop using seven high-quality datasets, containing over 358,000 images and 1,270,000 single cells with varying resolutions and channels from cellular organelles to tissues. Our results demonstrate Microsnoop's robustness and state-of-the-art performance in all biological applications, outperforming previous generalist and even custom algorithms. Furthermore, we presented its potential contribution for multi-modal studies. Microsnoop is highly inclusive of GPU and CPU capabilities, and can be freely and easily deployed on local or cloud computing platforms.

Minimum Minutes Machine-Learning Microfluidic Microbe Monitoring Method (M7)

“…Machine learning has greatly advanced biomedical research. − A flowchart of the machine-learning process is shown in Figure d. PCA is used to process spectral data with high dimensions, and by calculating the covariance matrix of spectral data, the correlation between different wavelength points can be analyzed.…”

Deep-Learning Terahertz Single-Cell Metabolic Viability Study

“…However, the cell samples should be placed in culture medium when evaluating cell viability using commonly used methods. Thus, according to the morphological characteristics of dead cells and living cells, a cell classification network (CCN) was built based on a convolutional neural network, a common machine learning and artificial intelligence method. ,− The cell viability assessment model based on CCN can be used to evaluate the viability of cell samples in nonliquid environments. Under the conditions of water stress, osmotic dehydration occurs in cells.…”