Science News

Deep learning allows us to identify blood cancer cells in milliseconds

A device capable of detecting cancer cells in seconds was developed by a group of researchers from the University of California at Los Angeles and NantWorks, a private US company.

The possibility of detecting cancer cells in the blood practically in real-time could allow them to be extracted in time, which would help prevent the spread of the disease.

The study, published in Nature Scientific Reports, explains how the technique that is based on two technologies works: artificial intelligence based on deep learning (deep learning), used to classify and generally analyze the data obtained, and time stretch (temporal extension) photonic.

The latter is an ultra-fast measurement technology invented by scientists from the Californian University that sees the use of ultra-short laser flashes to capture trillions of data points per second, a speed 1,000 times faster than today’s fastest microprocessors.

In addition to these two basic technologies, the method also uses a third technology called image flow cytometry. Cytometry is the science that measures the characteristics of cells and in flow cytometry of images these characteristics are measured by a laser for image acquisition while the cells themselves flow one at a time through a vector fluid.

Yueqin Li, a doctoral student and first author of the study, explains the system: “We optimized the design of the deep neural network to manage the large amounts of data created by our temporal extension cytometer, improving performance both the software and the tool.”