Slashdot Mirror
Massively Parallel X-Ray Holography
Roland Piquepaille writes "An international group of scientists has produced some of the sharpest x-ray holograms of microscopic objects ever made. According to one of them, they improved the efficiency of holography by a factor of 2,500. In order to achieve these spectacular results, they put a uniformly redundant array next to the object to image. And they found that this parallel approach multiplied 'the efficiency of X-ray Fourier transform holography by more than three orders of magnitude, approaching that of a perfect lens.' Besides these impressive achievements, it's worth noting that this technology has been inspired by the pinhole camera, a technique used by ancient Greeks. 'By knowing the precise layout of a pinhole array, including the different sizes of the different pinholes, a computer can recover a bright, high-resolution image numerically.'"
http://arxiv.org/PS_cache/arxiv/pdf/0801/0801.4969v3.pdf
It can't be used on living organisms as it causes instant cancer and death. Sorry, Bob.
Actually, from TFA:
So bright was the flash of light that the sample [bacteria] was vaporized, but not before both the scattered object beam and the reference beams from the URA had been recorded.
Indeed, as one can see, the part of "instant death" is quite accurate.
-><- no
The arxiv paper (referenced elsewhere in these posts) mentions that the obtainable information drops rapidly at 75nm. Their phase-recovery algorithm, combined with the snr inherent in the system, conspire to do this. It's really not a function of the computer post-processing (which can't, after all, improve the image resolution). The caption on one of the figures in the linked article is simply a little misleading; however, the entire article is quite good. Science reporting ftw!
i'm no expert, but i think by 'processing' they mean the fourier transform that is needed to get a 'usable' image from the hologram.
So, no interpolation, but a kind of signal processing, sort of like what your cell phone/wifi does to make sense a jumble of transmissions.
Check out:
http://www-group.slac.stanford.edu/ais/publicDocs/presentation71.pdf
Whatever information one can get is present in the original diffraction pattern. "Processing" *probably* means interpolation, or convolution with the known regular array. One can only keep the same information already present, or lose information in this way. They probably mean that the pattern was smoothed so as to look nicer to the eye (which is certainly valid), but I doubt they increased resolution in any way.
It's not that the computer processing improved the resolution, it's that the computer processing is a necessary part of the process which improves the resolution.
This article talks about taking normal x-ray radiation and using that to make a hologram. Holograms are usually made from laser illumination because a laser beam is coherent light, light in which the waves are all in phase (in step) with each other. It is difficult to make an x-ray laser but there is another way to get coherent light and that is through the use of pinholes.
The major problem of pinholes is that the smaller they are the less light is let through so the dimmer the image. However a large pinhole produces a very inaccurate (low resolution) image. One answer is to use a lot of small pinholes in the place of a few large pinholes. This is a great solution which produces sharp, bright images but now there is an additional problem, each pinhole makes a separate image and all these pinholes cause multiple overlapping images, offset a tiny bit from each other. This is where the computer works. Since the original pinhole pattern is known (you created it) you can feed that pattern into the computer and it can use that pattern to "slide" all the overlapping images so they exactly fit on each other. This makes a single, bright, sharp image.
The computer is not increasing the resolution of the detector, that's fixed. What it is doing is working as part of a process to produce better images.
Sapere aude!
Roland usually submits stories to Slashdot detailing some "breakthrough" scientific or technological discovery. Unfortunately, the links point to his ad-heavy blog and not to the story itself, thereby allowing Roland to get a sizable chunk of ad revenue courtesy of the hordes of Slashdotters who actually RTFA(ahem..). Note: 'usually'. Recently, his links have been pointing to the story directly, I'll give him that.
Greeks and Optics
The earliest known written evidence of a camera obscura can be found in Aristotle's documentation of a device in 350 BC in Problemata" (Patti, 1993). Aristotle's apparatus contained a dark chamber that had a single small hole to allow for sunlight to enter. With this device, he made observations of the sun. He noted that no matter what shape the hole was, it would still display the sun correctly as a round object. Another observation that he made was that when the distance between the aperture (the tiny hole) and the surface with the image increased, the image would become amplified. Although no one is perfectly sure, many attribute the invention of the camera obscura to Aristotle. He rejected the vision theory of Plato of light rays emitted from the eyes.
Camera Obscura
Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
Actually, after going back to the article and re-reading it I find that they are using pinholes to produce coherent reference light but they are only using two of them to do this, not a pattern of many of them. In the method described in the article the pattern is instead off to one side of the object to be imaged. It appears that they are using the pattern to deconvolute the final image. Since the pattern is known you can use a deconvolution function based on that pattern to re-create the original pattern in the image. This has the side effect of correcting the image of the object you wish to view, increasing the resolution of the image.
In essence they are using the pattern to calibrate their instrument in order to improve the imaging.
Sapere aude!
This is where the computer works. Since the original pinhole pattern is known (you created it) you can feed that pattern into the computer and it can use that pattern to "slide" all the overlapping images so they exactly fit on each other. This makes a single, bright, sharp image.
The computer not only deconvolutes the pinhole pattern(which only provides an image with the same resolution as the pinhole) but it uses the entire diffraction in the detector and phases it to obtain an image, much like a lens would do. This achieves a resolution that is simply limited by the numerical aperture of the detector (which can be much smaller than the pinhole size).
This is what a typical X-ray diffraction pattern looks like.
You need to get a sample like that from many angles of reflection, and then use Fourier transforms to piece it back together (which is why pictures wouldn't have been very interesting).
Also I noticed some saying "this can't be used on humans" etc. The idea is to use it to map out the internal structure of molecules, not to see people's bones.
e.g. DNA's discovery was based on this technique, and it's used constantly by the pharmaceutical companies to see whether their reactions are giving the chemicals they want.
So improved techniques like these are a big help to medicine.
(Only just starting a semester on this subject at the moment, let me know if I've missed anything)
// MD_Update(&m,buf,j);