Deepometry, a framework for applying supervised and weakly supervised deep learning to imaging cytometry

Abstract

This protocol describes Deepometry, an open-source application for supervised and weakly supervised deep learning analysis of imaging flow cytometry datasets. The protocol provides runtime scripts for Python, MATLAB and a stand-alone application.TWEET A new protocol for deep learning analysis of imaging flow cytometry datasets using #Deepometry. COVER TEASER Deep learning analysis of imaging cytometry data

Integrating Micro and Nano Technologies for Cell Engineering and Analysis: Toward the Next Generation of Cell Therapy Workflows

“…These sorters are also compact benchtop units that can eliminate sample handling challenges present with the usage of FACS in shared facilities. Finally, sorting based on imaging cytometry is gaining traction, , which when combined with deep learning enhanced analysis on these microfluidic platforms, can provide advanced sorting functionalities and high-dimensional information about cell morphology …”

“…Finally, sorting based on imaging cytometry is gaining traction, 52,53 which when combined with deep learning enhanced analysis on these microfluidic platforms, can provide advanced sorting functionalities and high-dimensional information about cell morphology. 54 Apart from microfluidic chip-based single cell sorting and isolation techniques, there have been advances in the design and implementation of miniaturized robots for the direct capture, transportation and manipulation of cells that have the potential for use in precision single cell biopsy from tissues or tumors. 55−57 One recent example is the use of untethered thermoresponsive grippers for the capture of single live cells or the extraction of a few cells from a clump.…”

scifAI: Explainable machine learning for profiling the immunological synapse and functional characterization of therapeutic antibodies

“…Recent studies have demonstrated the potential of machine learning algorithms for a more robust and accurate analysis of high-throughput imaging data, an approach that has been demonstrated to overcome limitations of conventional gating strategies [22][23][24] . Leveraging machine learning for IFC data analysis has also enabled the identification of morphological patterns in the cell, a combined analysis of RNA and protein data, and the implementation of predictive models [22][23][24][25][26] . While limited open-source software implementations designed for IFC data analysis are available 26,27 , they either rely on additional third-party software adding complexity in the analysis pipeline, or they focus on prediction performance only and lack explainability.…”

Self-supervised deep learning encodes high-resolution features of protein subcellular localization

“…In the case of supervised approaches, human annotators examine images and assign annotations, and once sufficient data are garnered, a machine learning model is trained in a supervised manner and later applied to unannotated data 17,18,23,24 . Another approach consists of reusing models trained on natural images to learn generic features on which supervised training can be bootstrapped 5,25,26 . While successful, these approaches suffer from potential biases, as manual annotation imposes our own preconceptions.…”