The technology firm Deepcell has developed an AI-powered single cell analysis solution, called the REM-I platform. The new platform enables individual cell imaging and sorting based on cellular morphology, offering applications to a range of fields including cancer, developmental biology, stem cell biology, gene therapy, and functional screening.
Deepcell is a company that spun out of Stanford and which seeks to understand cells via imaging and AI-analysis. These insights into cell biology are achieved through scalable single cell imaging, high-dimensional analysis and cell sorting
During the development phase, Deepcell technology has been used to capture and characterize more than two billion images of single cells across a large variety of cell types.
Cell morphology was one of the first ways cells were studied since the advent of the microscope. Despite recent advancements in microscopy and flow cytometry, existing tools for cellular quantification and characterization have left the field of cell biology hypothesis bounded and reliant on human interpretation.
This looks set to change with the new technology. With the new generation of AI and machine learning models cell morphology can join other high-dimensional, single cell analysis methods and enable researchers to realize the full potential of the morpholome.
The basis is the Human Foundation Model, a self-supervised deep learning model trained on a subset of these unlabelled cellular images from a range of carefully selected biological samples. This model characterizes brightfield single cell images captured on the REM-I instrument and generates high-dimensional embedding data.
Through this, researchers can use the Axon data suite to access, visualize, and analyze these data in real-time and perform sorting of their cell groups of interest into up to six outlets on the REM-I instrument.
With the help of sophisticated artificial intelligence models, researchers can surpass the limits of what their eyes can see and peer ever more deeply into the biology of individual cells. It is hoped that the REM-I Platform will catalyze new methods of discovery in a wide range of fields including cancer biology, developmental biology, stem cell biology, gene therapy and functional screening.
For example, advances in machine learning can transform our understanding of cell phenotype akin to the way next-generation sequencing transformed our understanding of the genome. In a second example, the Deepcell platform gives scientists the ability to discriminate between activated and naive T cells and provides next-level detection of therapy response in peripheral blood mononuclear cells derived from patients treated with immune therapies for cancer.
The research will be presented at CYTO 2023, highlighting deep learning capabilities of its artificial intelligence solution, the Human Foundation Model. The event is the Congress for the International Society for the Advancement of Cytometry conference.
