Talk:Phase-contrast X-ray imaging

Absorption cross section
I have been correcting minor things in this article, but when reaching the second half of the physical principle section I don't know what to do. It now says:


 * "Far from the absorption edges (peaks in the absorption cross-section due to the enhanced probability for the absorption of a photon that has a frequency close to the resonance frequency of the medium), dispersion effects can be neglected; this is the case for light elements (atomic number Z<40) that are the components of human tissue and X-ray energies above 20 keV, which are typically used in medical imaging.
 * Assuming these conditions, the absorption cross section is approximately stated by


 * $$\sigma_a=0.02[\text{barn}]\left (\frac{k_0}{k}\right )^3Z^4$$


 * where 0.02 is a constant given in barn, the typical unit of particle interaction cross section area, $k$ the length of the wave vector, $k _{0}$ the length of a wave vector with wavelength of 1 Angstrom and $Z$ the atomic number. "

I don't see how the photo-electric absorption cross section is relevant, when Compton scattering has a higher cross section in tissue above 25 keV. Compton scattering is relatively independent of photon energy which makes the advantage of phase contrast decrease when increasing the energy (at energies above 35 keV). Ulflund (talk) 12:43, 24 July 2013 (UTC)


 * Well, that comes directly from equation 2.3 of Martin Bech's Ph.D thesis, and Dr. Bech's profile, including what I presume is a current email address, is visible on this page: http://www.e17.ph.tum.de/index.php/people/14-martin-bech.html. I would just write to him to ask him about it. The direct approach works wonders. Stigmatella aurantiaca (talk) 13:12, 24 July 2013 (UTC)

Edge-illumination section
Whoever wrote the section about Edge-illumination technique seems to be quite impressed about his almighty technique. It reads like an advertising brochure for the method. Here are some points:

This plurality of individual beamlets means that no scanning is required – the sample is placed downstream of the sample mask and imaged in a single shot (two if phase retrieval is performed[22])
 * I don’t get the reasoning here, there are also single-shot techniques with GI systems. What makes it only possible with EI and what is so special about that?
 * This is simply a statement of fact, whether or not that makes EI "special" is therefore not relevant.

Prmunro (talk) 16:00, 9 September 2020 (UTC) In addition to edits made to the article, I would like to make the following points: (i) This is simply with respect to the original synchrotron implementation shown above (ii) This plurality of individual beamlets means that, compared to the synchrotron implementation discussed above, no scanning is required

''GI is an intrinsically coherent method, in which an incoherent source can be used only provided it is made sufficiently coherent through collimation via the source grating. In contrast, EI is an incoherent technique, and was in fact proven to work with both spatially and temporally incoherent sources, without any additional source aperturing or collimation.[22][82]''


 * Well, it is not at all mentioned that EI also needs some degree of spacial coherence of the source due to smearing of the beamlets created by the mask - and it’s even in the abstract of [22]. The source spot size has to be below 100 microns, as in almost all the papers of EI systems. That means the method is not compatible with common sources (angiography tubes, CT tubes) used in medicine.

Prmunro (talk) 16:00, 9 September 2020 (UTC) In addition to edits made to the article, I would like to make the following points: (i) The transverse coherence length at the pre-sample mask is sub-micron, while the mask period is typically between 50 and 100 micron: this 2-order-of-magnitude-difference justifies the incoherence claim. We also note that 100 micron focal spots are used clinically in e.g. diagnostic mammography. (ii) EI is an incoherent technique (e.g. the coherence length at the pre-sample mask is typically sub-micron, 2 orders of magnitude smaller than the mask period), and was in fact proven to work with both spatially and temporally incoherent sources, without any additional source aperturing or collimation.

The highly simplified set-up, which however does not lead to reduced phase sensitivity,[83] results in a number of advantages, which include reduced exposure time for the same source power, reduced radiation dose, robustness against environmental vibrations, and easier access to high X-ray energy.[83][84][85] 


 * It is completely not clear to what the EI technique is compared here. Is there a paper that clearly compares a Talbot-Lau Interferometer with a EI system of comparable size? Is there a clear comparison of both systems in terms of source power, dose and energy range? I don’t see a real rigorous quantitative comparison in the cited papers to support it the way it is written here. It's just some vague claims put together on some catchwords from the publications.

More generally, while in its simplest implementation beamlets match individual pixel rows (or pixels), the method is highly flexible, and, for example, sparse detectors and asymmetric masks can be used.[87]


 * Well, highly flexible, really??? The clear limit is the distance between both masks since the convolution of the source spot with the first mask magnified onto the detector mask has to be in the range of about 20 microns. That means EI is not flexible at all. The length of the system has to be around 2m and there is only 40-60cm place for the imaged object. Hence, it cannot be implemented within a clinical CT scanner or other clinical applications. Where is the clear reasoning e.g. in a publication that EI is more flexible than GI in terms of length, focal spot size and sample FOV?

Prmunro (talk) 16:00, 9 September 2020 (UTC) I am unsure as to where the assertion that the system length is fixed at 2m originates from. For example our current intra-operative system is 85 cm long, and systems with a shorter overall length and varying source-to-sample and sample-to-detector distances have been published already several years ago (e.g. Endrizzi et al. Opt. Lett. 39 (2014) 3332-5). We have amended the text to make this clearer

A new chapter "Applications"
The article should have an additional section showing the applications of DPC imaging. Up to now, there are applications mentioned scattered all around the article. In most cases they are not specifically possible with a certain technique, therefore they can be treated separately in a new chapter.