Cancer cells that break away from a tumor and metastasize lead to 90 percent of all cancer deaths. Researchers have spent decades trying to develop blood tests that can effectively detect these circulating tumor cells. Finding them, however, can be like searching for a particular needle in a stack of needles. One milliliter of blood contains about five billion red blood cells and nearly 10 million white blood cells but only 10 tumor cells.

Researchers at the University of California, Los Angeles, have developed specialized technology that may be able to find these cells before they form new tumors, significantly boosting a patient's odds of survival. They describe the system in a July issue of the Proceedings of the National Academy of Sciences USA.

At the heart of the U.C.L.A. system is an ultrafast microscopic camera the researchers introduced in 2009 that captures images at about six million frames per second. This so-called serial time-encoded amplified microscopy (STEAM) camera creates each image using a very short laser pulse—a flash of light only a billionth of a second long. Its shutter speed is 27 picoseconds, about a million times faster than a current digital camera. (A picosecond is one trillionth of a second.)

The U.C.L.A. camera converts each laser pulse into a data stream from which a high-speed image can be assembled. To the STEAM camera, the investigators have added a microfluidic channel for the cells to flow through and a high-speed image processor that, they say, takes blur-free images. The team used this technology to identify breast cancer cells in blood samples. “We look at a cell's shape, size and texture, as well as its surface biochemistry,” explains lead author Keisuke Goda, who recently moved from U.C.L.A. to the University of Tokyo. “Cancer cells tend to be larger than white or red blood cells. And we know that a cancer cell's shape is ill defined compared with red and white blood cells.” Goda adds that a relatively noninvasive blood test would encourage people to get screened more frequently than they do now.

COMMENT AT ScientificAmerican.com/oct2012