Okay, well, not exactly. But pretty freaking close. A "just accepted" article in the journal Analytical Chemistry from Aydogan Ozcan's group at UCLA describes a cellphone-based device for detecting and counting fluorescence-labeled cells (or other fluorescing particles) flowing past the cellphone camera.
This device fantastically uses LEDs pointing in at each other from each end of a cheap microfluidic chamber (i.e. a silicone-coated glass chip) that acts as both the channel for the cells to flow through and an optical waveguide (making the light waves from the LED oscillate and bounce around, kind of like how a fiberoptic works) that enables fluorescence excitation and emission, which gets sent through some cheap lenses and an optical filter (to isolate the fluorescence signal) and picked up by the cellphone's integrated camera.
To demonstrate that this could give useful biological information, they combined the device and phone with an external pump to flow liquid containing particles (either a suspension of fluorescent particles or diluted blood) past the camera. In the version where they analyzed blood, the blood had been dyed with a fluorescent stain that made the white blood cells glow in the LED setup (because the dye stains DNA, and red blood cells don't contain any DNA). Then they wrote some software that post-processed the movie images and counted the particles or cells.They were able to get pretty good detection and resolution of particles as small as 1 micrometer in diameter--which is smaller than most human cells (meaning that even smaller things, like bacteria, might be able to be detected and distinguished from each other).
Could this really be used to diagnose cancer? Sort of. White blood cell counts are a key metric by which leukemia and lymphoma (cancers of the blood) are diagnosed and monitored. This device was able to get white blood cell counts that were really close to those obtained with a fancy blood cell counter instrument. Of course, to do anything more than just count the number of easily stainable white blood cells, people with a lot of biomedical expertise will have to think of ways to apply this detector. For example, from the proof-of-concept experiments they showed, you wouldn't be able to tell what kind of blood cancer someone had, and you wouldn't even get very much information about the status of their blood cancer if they had started on a treatment. Even so, being able to do this accurately with something that costs $50 is revolutionary: most fluorescence-based detectors cost tens of thousands of dollars and require sensitive, fancy optics to work properly. Even the "cheap" versions are a few thousand dollars. This is some LEDS stuck onto a glass chunk with some plastic bits attached. And it works.
Why is this so amazing? Well, apart from that it is small and readable by an extremely common and widely available camera, it is also made of all low-cost parts, so it will be extremely accessible yet gives pretty sophisticated information (fluorescence image readings). Its applications are only limited to what kinds of dyes are available to specifically stain different things you might want to look for in cells or particles, and the ingenuity of the applier in figuring out what to stain with fluorescence and try to look for. There are fluorescence-type stains that can be applied to cancer cells to figure out more about the type, stage and response to treatment of the disease--and while not trivial to apply with this device, potentially within reasonable optimization reach. They can even use different cheapo plastic optical filters to look for different fluorescence colors (besides just the green fluorescence they show in the paper). Of course, if they want to try to look for more than one color at a time (called 'multiplexed' fluorescence), things might get complicated.That requires more than one LED and more than one optical filter, but, I imagine, isn't impossible.
Beyond just white blood cell counts, I can envision this being used as a readout technique for detecting circulating tumor cells (CTCs)--rare cells that are shed by other kinds of tumors into the blood stream. CTCs usually have different proteins and carbohydrates on their outer surfaces than blood cells do, so it's possible to either stain them with fluorescent dyes that don't stain the blood cells, or catch them using antibody grabbers in microfluidic chambers (very similar to the one used in this device) that latch onto those specific proteins and carbohydrates, but let all the other cells flow by. In that case, you could then just stain the captured cells with the same antibody grabbers (left free floating rather than attached to the chamber, and pre-labeled with a fluorescent dye) or some other generic dye for staining all cells or cell parts (like the nucleus).
It will need to be integrated with some kind of pump to be more universally useful, because not everybody has a syringe pump sitting at the back of a closet that they can just hook up and go. It'll also probably need some integrated sample well that can keep the sample 'agitated' (i.e. shaking) during the flow so the particles don't settle and clump up. However, these should be easy enough to figure out (maybe they can plug into the vibration unit on the phone to do the shaking?). They designed their counting software using a platform that is compatible with Android phones--so even though the phones they used in this study couldn't do any integrated processing, it's a very short jump to adapt this device to an Android phone or iPhone and have... (wait for it...) an app for that. lol. I am fascinated to see how this device gets further developed and used, and I plan to buy one for myself as soon as it's available.