Scientists Hack Cellphone To Detect Diseases

Dave Bullock (eecue) plugs his piece up at Wired on a cellphone modded into a portable blood tester. This could become a significant piece of medical technology. "A new MacGyver-esque cellphone hack could bring cheap, on-the-spot disease detection to even the most remote villages on the planet. Using only an LED, plastic light filter, and some wires, scientists at UCLA have modded a cellphone into a portable blood tester capable of detecting HIV, malaria, and other illnesses. Blood tests today require either refrigerator-sized machines that cost hundreds of thousands of dollars or a trained technician who manually identifies and counts cells under a microscope. These systems are slow, expensive and require dedicated labs to function. And soon they could be a thing of the past."

How it works

[marcansoft]

If anyone is wondering how exactly this works, or wants to build their own, they might want to check out this Weekend Project [youtube.com]

Basically, if you've got a (near-)point source of light, and transmit it through the sample, there is only one path of light from the light, through a point in the sample, to a pixel on the sensor, so you don't need a lens. The farther away you place the sample and the closer you place the light source, the larger the image appears (but then you also need to progressively use a better, closer to a true point source light).

I imagine this could work very well with a naked silicon laser diode, since they appear as damn tiny, near point sources of light.

Bullshit

[toppavak]

A. It can't detect HIV. No imaging technology short of electron microscopy can directly detect the virus itself and even electron microscope would be a retarded way to attempt diagnostics.

Even the original paper describing this technology showed that they have no sub-cellular resolution and even their size resolution was extremely unreliable for anything smaller than 15 microns... which all interesting human cells are (even if you could tell what size cells are you've accomplished.... nothing).

If they are suggesting they can do CD4+ T-lymphocyte counts they're either idiots, ignorant or both. There is no morphological distinction between a CD4+ and a CD4- T-lymphocyte. Even using fluorescence imaging (which they aren't) you have to be able to look at two colors of fluorescence (CD3 label to check to see if its a lymphocyte and a CD4 label to see if its CD4+) immunofluorescence is way too weak to be detected by a cell phone camera, especially a color sensor with 2 micron pixels. The CD4 antigen is never expressed at levels greater than approximately 50,000 / cell, the detection limit of a 5 micron pixel monochrome sensor (the bayer mask makes you lose about 30% of your light) is close to about 150,000 molecules. The bayer mask also makes your sensor pretty much useless for analytical applications, you're screwed if your green-fluorescent cell is centered over a red or blue-sensitive pixel which would happen in, oh, 66% of your pixels.

You run into almost identical sets of problems with every other so-called "application" of this "technology" so, yeah, bullshit.

IAABME.

Re:How it works

[toppavak]

To elaborate on the inadequacy of this technique for malaria diagnostics, for example, the Poisson statistics indicate that a reliable diagnosis requires the analysis of about 2 microliters of blood. Now this doesn't seem like much except that 2 microliters of blood contains 10-14 million red blood cells. In order to just have all of those cells on the sensor while able to discriminate between adjacent cells that means you'd need at least a 90-megapixel camera on which to smear your blood sample. Now if you wanted to reliably detect a malaria infection, you would obviously need a lot more than 1 pixel per cell so the magnitude of the problem increases exponentially from there.

Re:star trek isn't dead yet

[Spiked_Three]

"Now, if Bill G were really interested in changing the world's health... perhaps he'd get on-board with this obvious idea. Who knows. He's got a lot of money."

For some reason, I was bored and just happen to watch the PowerPoint presentation on this thing last week on TV from Washington University. It was quite an impressive device, with many possibilities for multiple uses as part of the design criteria.
 
And the Bill & Melinda Gates foundation was a huge contributor.

maybe you shouldn't assume so much

[boombaard]

Your "Facts" are wrong on so many levels it isn't even funny.
Please have a look at this pdf [pnrec.org] (which admittedly uses OECD data from AD 2000, so they might be somewhat outdated, but it will do to make my point)

On page 9 you'll see that public health spending (as a %age of total) is lowest in the US, (and highest in Sweden) and on page 10 you'll see that the total amount spent per person on healthcare in the US is nearly 73% higher than in the next country listed (Germany).
Next, if you have a look at the CIA World Factbook: (website isn't working here, so using wikisource [wikisource.org])
and look at the figures for average life expectancy in the US compared to socialist Europe, the average in the US: about 74 (male) and 80 (female), whereas in Sweden, (which admittedly has better life expectancy than some EU countries, but i can't be arsed to find the median country) it's 78/82 years (2004 est)).
Additionally, the Infant mortality rates:
US: 6.63 deaths/1,000 live births
Swe: total: 2.77 deaths/1,000 live births

Sweden's per capita spending: less than 1700$
United States per capita spending: 4100$

Please show me how or why "government healthcare is bound to fail", or, alternatively, have a look at actual data.

(Disclaimer: since 2004 a number of european countries are reforming/considering reforms to health care funding, because it's inefficient in some ways. Nevertheless, the fact remains that health care spending here costs less than half of what it costs in the US.)