Talk:Diffusion-weighted magnetic resonance imaging

Merging

"a tensor (i.e. a symmetric positive definite 3 ×3 matrix)"

Diffusion tensors are positive definite, but this does not mean that all tensors are.

Merging Diffusion MRI and this article is not reasonable, because diffusion MRI relates to the measurement itself, but Diffusion-Tensor-MRI only to a special method of evaluating and analyzing the data, i.e. by mapping it to a second order symmetric tensor. There are a lot of other techniques such as q-Ball imaging (by David Solomon Tuch), q-Space imaging and other methods that are summed up by high angular resolution diffusion imaging (HARDI) that either depend on higher order tensors or other representations (such as spherical harmonics, that can be seen in this very special case a tensor represented in a spherical coordinate system - or other techniques, that base on discrete samplings of the sphere, exist, too.)

So there should be one article explaining the basics of diffusion imaging (compared to simple MRI) and one or more articles describing the methods of diffusion tensors, higher order tensors, HARDI, q-Space and qBall methods. So the split should be inbetween the measurement of the data and the analysis.

While I completely agree with you that Diffusion MRI and DT-MRI are by no means synonymous, my point is that the current WP-entries do not reflect the distinction sufficiently to warrant two different articles. At least 80% of the current article on Diffusion MRI repeats content from Diffusion Tensor MRI. As long as the articles are not more verbose, I don't see a problem in merging them, explicitly explaining the difference between the terms and letting one term redirect to the other (keep Diffusion MRI, if you wish — DTI is just a subset of it). If at some point the article should become too long, one may split it again in the way you suggest. --Thomas Schultz 16:38, 2 February 2007 (UTC)

I agree

It seems the two are close enough and should be under a common title, preferably Diffusion MRI, as long as searching the terms DWI and DTI (and their respective expansions) link one to Diffusion MRI. That entry was more descriptive and conatined more useful information for a "dumb" GP, non-radiologist type.

P.S. Dr Schultz, that image is quite beautiful (or, as a teen would say "awesome") Mlintx 21:52, 6 July 2007 (UTC) Martin Lindenberg

I think they should be merged, there is simply not enough content to justify separate articles. They are separate things but until the DTI article becomes complicated enough it does justify its own article (the same as the single paragraph about MRA under the MRI article doesn't need its own article) --SBarnes 3 August 2007

Phrasing

I just noticed the following:

"This causes the millions of water molecules to precess simultaneously, and it is this precession of protons which produces signals in MRI."

For clarity, I am changing it to:

"This causes many of the protons in water molecules to precess simultaneously, producing signals in MRI."

-- J. Hengenius (grumpy_mr_gruff@yahoo.com) 10 March 2008 —Preceding unsigned comment added by 128.210.89.66 (talk) 15:57, 10 March 2008 (UTC)

History

when was this technology developed? It seems pretty new (and spectacular!!). --Shaggorama (talk) 08:58, 1 May 2008 (UTC)

Thanks - a history section and a copy of the first DTI image from 1992 has been added - then supplemented with additional references. (Afiller (talk) 04:50, 29 May 2009 (UTC))

There should be several subsections here. One for each of the areas: diffusion MRI, diffusion vector imaging (DVI) of Filler and coworkers, diffusion tensor imaging, HARDI, Q-Ball, Generalized DTI, Kurtosis and others. —Preceding unsigned comment added by BasicTruth (talk • contribs) 14:47, 22 July 2009 (UTC)

Basic Truth and Afiller---It is a pity that the history section has caused so much bad feelings on Wiki. Wouldn't it be nice if both of you contribute to writing the history of diffusion MRI, diffusion vector imaging and diffusion tensor imaging, which were invented/proposed in the 80's and early 90's. Other fancier techniques invented in the late 90's and this century may be put elsewhere. What do both of you think about this suggestion? CogitoErgoSum101 (talk) 15:16, 22 July 2009 (UTC)

Please note that "Basic Truth" has been editing my discussion comments. There is no "DVI" or "diffusion vector imaging" as a subfield. It was the first successful tractographic image. Is there any rule on Wikipedia against someone else editing comments that I have signed.?Afiller (talk) 22:15, 22 July 2009 (UTC)

CogitoErgoSum101 - Thanks for the suggestion. Earlier, various editors combined diffusion weighted imaging and diffusion tensor imaging into a single article and I do think that makes sense. I think Basic Truth's editing and personal attacks will deter some from writing more subsections on this article - what if he disagrees with something someone writes?, but we'll see. Perhaps he will fill in his references and continue writing usefully. Afiller (talk) 23:30, 22 July 2009 (UTC)

CogitoErgoSum101- Oh well, I guess I'm getting used to the thrashings by 'Basic Truth' so I'll try to add those additional sections and be careful to save my work frequently.Afiller (talk) 05:05, 23 July 2009 (UTC)

CogitoErgoSum101 - I have added HARDI and Q-Ball sections to the history. I don't think this makes the article overly long. Hopefully "Basic Truth" won't trash it. In the future, other experts in HARDI and Q-Ball could always add and polish if needed.Afiller (talk) 07:24, 23 July 2009 (UTC)

The History subsection of the section "Diffusion Imaging" seems like a bit of a tribute to Le Bihan. Almost as if someone from his lab wrote it. For example, take the line "it worked beautifully and that move was a great achievement". Doesn't sound very objective. — Preceding unsigned comment added by Monsterman222 (talk • contribs) 20:38, 24 October 2012 (UTC)

accessibility

This article needs a better introduction that gives a more basic explanation of the concept. I'm quite scientifically literate and this article leaves me scratching my head. —Preceding unsigned comment added by 76.181.70.75 (talk) 04:13, 6 April 2009 (UTC)

A new introductory section has been added (Afiller (talk) 04:51, 29 May 2009 (UTC))

It would likely be helpful to define the variables used in the equations. Even someone sufficiently mathematically literate to understand the equations might not understand their applications because the relationships of the variables to the physical processes are not explicitly stated.(JHUbme24 (talk) 15:34, 28 August 2009 (UTC -5:00))

Photo License

I know this is tedious, but we need to get it right. The photo of Aaron Filler is licensed CC ShareAlike 3.0, but the summary contains the following restriction: Permission is granted for use in educational, research or academic publications or news sources media.. Is this acceptable for Wikipedia? It appears to invalidate the ShareAlike license (I am no expert in this area!). I think the idea is that any Wikipedia page, complete with illustrations, can be reproduced anywhere by anyone. Does anyone have any more specific information? GyroMagician (talk) 09:35, 21 July 2009 (UTC)

GyroMagician - thanks. I would certainly adjust the rights as required.Afiller (talk) 01:30, 22 July 2009 (UTC)

No problem ;-) The images have currently gone - was that you? It's a shame to lose them, I like to have a few images showing the original work actually being done, and it's relatively rare to be able to put them on Wikipedia (license and all). GyroMagician (talk) 12:05, 22 July 2009 (UTC)

GyroMagician - no that's "Basic Truth". Thanks. It's fascinating to see him struggle with the history he'd like to delete. Eventually - as I said - he will erase all the information about our group. Fortunately, he can't erase patent literature and academic publications - much as I'm sure he'd like to. Oh well, -as the age old expression goes - if you can't take the heat, you ought to stay out of the blogosphere. I know that certainly applies to Wikipedia as well.Afiller (talk) 16:52, 22 July 2009 (UTC)

DVI vs. DTI

I have moved the discussion from the top of the page down to here, to try to keep the talk page readable. At the same time I put a heading on the merge discussion, for the same reason. Comments are otherwise unchanged. GyroMagician (talk) 19:15, 21 July 2009 (UTC)

There is no doubt that there should be four different wiki pages: diffusion MRI, diffusion vector imaging (DVI) for Dr. Aaron Filler to take care of, and diffusion tensor imaging (DTI) for the DTIers to chip in and high angular resolution diffusion imaging (HARDI) for HARDI followers to contribute. The history section is disgustingly distorted. —Preceding unsigned comment added by BasicTruth (talk • contribs) 02:32, 21 July 2009 (UTC)

Basic truth - the problem with "DVI" is that it this is a public encyclopedia, not a debating society. This is a new term you are proposing here. The tensor method was well disclosed in 1965 for NMR if you will read the references. The solution of generating tractographic images from diffusion data is the basis of DTI this comes from the Filler 1992 reference even if that is upsetting to you personally. The same document in which the tractographic discovery was published describes the acquisition of diffusion gradients in an infinite number of directions for tensor analysis. The historical fact is that this was the first image that demonstrated how diffusion method could be used for tractography. The diffusion tensor ellipsoid model is not applicable to HARDI or Q-Ball methods. As the article points out, once Dr. Moseley explained to 700 MRI scientists how to do tractography from vector and tensor methods, in March of 1993, numerous scientists began developing improvements. Peter Basser filed his patent 5 months after Moseley presented the work by Filler & Richards. Moseley himself pointed out the tensor basis of the data in 1990. Afiller (talk) 15:14, 21 July 2009 (UTC)

Afiller - this is a public encyclopedia, not a place for you to distort history. Facts are facts. Three of your (co)authored papers in dated from 1992 -1993 did not even mention the term term "tensor". Here they are: [1] Title: MAGNETIC-RESONANCE NEUROGRAPHY Author(s): FILLER AG, HOWE FA, HAYES CE, et al. Source: LANCET Volume: 341 Issue: 8846 Pages: 659-661 Published: MAR 13 1993 [2] Title: MAGNETIC-RESONANCE NEUROGRAPHY Author(s): HOWE FA, FILLER AG, BELL BA, et al. Source: MAGNETIC RESONANCE IN MEDICINE Volume: 28 Issue: 2 Pages: 328-338 Published: DEC 1992 [3] Title: IMAGE NEUROGRAPHY ON STANDARD-GRADIENT MR-IMAGERS Author(s): FILLER AG, HOWE FA, WINN HR, et al. Source: RADIOLOGY Volume: 185 Pages: 152-152 Supplement: Suppl. S Published: NOV 1992

Facts are facts---please don't try to make a spin on history. As you mentioned, vector is a tensor of rank one. Then, LeBihan, Moseley and others could have claimed that they too had discovered diffusion tensor imaging because a diffusion coefficient is a scalar, which is a tensor of rank zero. But, diffusion tensor imaging as viewed by many in the community is specific to tensor of rank two. Huh..., unfortunately, you and your colleagues' Diffusion Vector Imaging is of "one rank" below DTI. —Preceding unsigned comment added by BasicTruth (talk • contribs) 19:26, 21 July 2009 (UTC)

(Basic Truth - we both edited at the same time, the following was written without reading your last edit)

I think you are both close to an 'edit war', which is generally bad for you (it's really not worth getting angry about) and bad for the page - please try to cool off a bit. I don't see a case for four different pages, unless there is enough material written to fill them. Keep it all on one page, we can decide to split it later if there is need. As a coil-guy, it's all just magnetic resonance to me anyway ;-)

Basic Truth - I must admit I have never heard of DVI, but am familiar with DTI (in name at least). That may well be because this is not directly my field - do you have references to back up the name (and ideally explain the difference)? You both seem to agree that HARDI exists, so why not add three sections explaining what Diffusion MRI, DTI and HARDI are (and maybe a fourth if DVI is included)? Try to aim them at someone with a basic grasp of MRI, rather than a specialist. And remember, no original research. GyroMagician (talk) 19:33, 21 July 2009 (UTC)

Basic Truth - I can see that you won't be comfortable unless you remove all reference to my group's work and contributions. If you think that makes things "unbiased" then I can see the way you function as an academic. Don't worry I won't waste my time with an "editing war." I recommend you read the references as this is all explained. The tensor reference appears in the correct place - in the July 1992 and March 1993 patent filings as cited. We originally called this "Diffusion Anisotropy Imaging" because it incorporated the various methods, but that is not the standard term.Afiller (talk) 20:18, 21 July 2009 (UTC)

Basic Truth - another concern is that although Basser et al called this "Diffusion Tensor Imaging" they were unable to produce an image - just mathematica style ellipsoid graphs. They would therefore have been more correct to call it "Diffusion Tensor Mathematica Graphs". To avoid going back and forth I changed "first DTI" to first diffusion tractographic image as I have not yet heard why this is incorrect. The point about the tensors is that - yes Edward Stejskal described the tensor ellipsoid formalism for NMR of diffusion in 1965 (not the more famous Stejskal & Tanner paper - see the reference in the article). Also, Dr. Moseley stated in his 1990 paper that the solution for anisotropy imaging was a tensor - one of the co-authors on that paper is a co-inventor on the Filler group patent. As far as we knew at the time, the use of the tensor ellipsoid formalism in diffusion anisotropy NMR had been well understood for 25 years so it didn't seem like something Peter Basser could re-invent.Afiller (talk) 23:12, 21 July 2009 (UTC)

Afiller - The term, "Diffusion Tensor", is well known in NMR and was described by Stejskal but how to come up with the acquisition design, and the methodology to compute this quantity---an inverse problem---is not known in the field MRI before a series of papers by Basser and coworkers. The situation is best described by the following passage by my favorite character, Sherlock Holmes, "In solving a problem of this sort, the grand thing is to be able to reason backward. That is a very useful accomplishment, and a very easy one, but people do not practise it much. In the everyday affairs of life it is more useful to reason forward, and so the other comes to be neglected. There are fifty who can reason synthetically for one who can reason analytically." Perhaps, there were many who thought that diffusion processes should be described by the diffusion tensor before 1992. But, the fact remains that none, at the time, had the know-how to "reason backward". Another point is that whatever methodology one is proposing, it should be enabling; it was the case with DTI.

In regards to your "new term" for the figures. You are trying to be trendy---please...just call it by the name you put it in your paper. Did you mention "tractography" in 1992 patents as well? Neurography sounds nice.

Basic Truth - We did originally call brain tractography "Neurography" although current usage is to apply neurography for peripheral nerves and "tractography" for neural tracts. However, just as DTI includes HARDI and Q-Ball that do not use the tensor formalism, this is the basis of DTI tractography as described in in the Filler et al references. The methodology for processing the data was well known from many many sources in 1990. Any chance you want to reveal who you are - just in case you have a conflict of interest?Afiller (talk) 00:24, 22 July 2009 (UTC)

Basic Truth - I see you are still applying edits - but how exactly do you prove that when the Filler group's work on how to do tractography was presented by Moseley at the SMRI plenary session in 1993 it had no effect, but that it was the 1994 paper by Basser that led to tractography (which he doesn't even mention in his paper or patent) when that field developed further in 1999 and 2000?Afiller (talk) 01:37, 22 July 2009 (UTC)

Afiller - DTI did not include HARDI and Q-ball. Generalized DTI (ala CL Liu and E Ozarslan), HARDI (ala Tuch, Frank, Alexander and many others) and Q-ball (Tuch) are different methodologies for describing (or modelling) diffusion anisotropy. DTI-based tractography started some time in 1998-2000 with the work of Mori et al., Conturo et al. and Basser et al.. Of course, there are many papers (Westin's and Alexander's groups to name a few) on this topic since then and some are HARDI or Q-ball based techniques. I am sorry to see that neurography did not catch fire then and I am confident to say that it is unlikely to catch fire now because there are many more advanced methodologies even compared to DTI. I don't have any conflict of interest and cannot stand to see this page being distorted for personal fame or gain. It is not a matter of who being the first or who being the first to get credit, what matters is getting things done and getting to the truth about brain science. Just move on, my man. And, continue your good work. —Preceding unsigned comment added by BasicTruth (talk • contribs) 03:52, 22 July 2009 (UTC)

Basic Truth - It's good to see that you have edited yourself in a circle without stopping to read the review or references. One of the inventors in the Filler group patent was part of most of the Alexander tractography research. Thanks for giving us the credit for tractography development. Neurography, by the way, is indeed a very active field with 10's of thousands of imaging studies per year. Courses are taught at RSNA, UCSF, Hopkins, etc once again you are embarrassing yourself while energetically wrecking the article because you have some sort of intense personal dislike for our publications and widely recognized contributions. Have you ever done anything to create a new discipline in imaging? - It appears you are a wrecker and not a builder.Afiller (talk) 04:07, 22 July 2009 (UTC)

It is laughable to say that the Alexander tractography research originated from your work. Just check the citation!!! Oops, they did not cite your papers. Simply google Mori et al., Conturo et al. and Alexander et al. and compare the number of citations of these works and yours. Yours lag by several hundreds!!! My god, your face is as thick as the Great Wall of China. —Preceding unsigned comment added by BasicTruth (talk • contribs) 04:25, 22 July 2009 (UTC)

Basic Truth - Thanks for the high quality comments. Most of my papers are in peripheral nerve neurography and various other fields as well as numerous patents and a number of books. You have written in the Wikipedia article that our work had no influence, but you misunderstood my previous paragraph in the discussion - one of our coinventors is a co-author on those Alexander papers - so how do you prove that his contributions to tractography in our patent had no effect on his contributions to tractography as a co-author with Alexander? I know this exchange is pointless. If you had the courage of your convictions you would identify yourself. No surprise but I'm guessing you'll stay anonymous.Afiller (talk) 04:34, 22 July 2009 (UTC)

Afiller-Your works are not yours alone but your coworkers as well. Perhaps, your coauthors formulated the Vector formulation? I am in no position to speculate. Coauthorship is a delicate issue as you might have learnt from your own experience. I hope more journal publishers will require authors to explicitly state their own contribution to the work in the paper. PNAS and many higher impact journals do. —Preceding unsigned comment added by BasicTruth (talk • contribs) 04:43, 22 July 2009 (UTC)

Afiller-I am sure those (Daniel Alexander or Andrew Alexander) papers are not based on your methodology but DTI. Just move on. Take a hike or ride a bike. —Preceding unsigned comment added by BasicTruth (talk • contribs) 04:56, 22 July 2009 (UTC)

Basic Truth - You are obviously an angry man on a mission, bent on anonymity in your work. I don't really see how deleting the first tractographic image helps the reader understand the history of DTI. But the important thing here is that it all make sense to you personally and fits closely with your pre-received world view. This is much the way in which religious fanatics proceed. You may be interested to know that even in the discussion section, Wikipedia encourages civility - perhaps you should revise that guideline for them.Afiller (talk) 07:59, 22 July 2009 (UTC)

Basic Truth - Susumu Mori - in his patent (which preceded his papers) the Filler patent is cited. The patent office does not allow what you would like - erase the prior contributions and inventions then claim they have nothing to do with a new novel idea. Mori - who is a wonderful and brilliant scientist and teacher - had to take advantage of the previous tractography work and build on it. The "DVI" story as you tell it is nonsense - the Filler group patents -e.g. the July 31, 1992 UK patent fully covers the tensor methods. The vector image solution just came first historically and was part of our work. There is no field of DVI - it represents a few weeks work in a single project whose publication includes the vector and tensor methods. It reveals how you obtain a rotationally invariant solution, produce a single image out of multiple gradient images, calculate Euler angles, then do progressive selective tractography based on anisotropy and directional information. We went on to explain combining mulitple gradients to achieve an infinite number of other gradients and using them in tensor solutions, then applying a connected voxel algorithm. Before this, groups were publishing three single images or crude color mixes. You can erase Wikipedia pages, but you must find it very frustrating that you can't erase history.Afiller (talk) 14:13, 22 July 2009 (UTC)

Afiller-Prior work should of course be cited in patent applications but the question is whether your "technology" is enabling. I don't think it is. Have you filed law suits against infringement by any of the big medical companies? I guess not. —Preceding unsigned comment added by BasicTruth (talk • contribs) 14:44, 22 July 2009 (UTC)

Basic Truth - The same issue applies to Pierpaoli. There we are cited more than 30 times in the file history (published and publicly available) for a patent filing by Pierpaoli and Basser as the examiner repeatedly points out to them that they cannot reinvent what we have already invented - and you certainly cannot claim that they were unaware and uninfluenced. Your solution that I personally and simultaneously - as an individual - sue GE, Siemens, Philips and Toshiba, etc is charming since each suit is 5 to 20 million to prosecute. Nonetheless the answer is that these companies do take our patents very seriously and are well aware. Counsel for Philips in one of their DTI patents where we are cited more than 30 times in the file history clearly accepts that we invented DTI and tractography but that they have some improvements. When you look at Hounsfield, the material is in the patents - show me a paper he published that explains in detail how to build and operate a CT scanner. What's your point? The fact is that we did publish, our work was re-presented by Moseley at a plenary session, all four of us continued to publish vigorously in this and other areas, the work has been covered by the New York Times, London Times, BBC, Peter Jennings ABC News and appears in prominent journals like Lancet. You know what - you are just wrong and really ought to be apologizing for some of the language - or stop hiding behind anonymity and take responsibility for your embarassing lack of knowledge of your own field. It is particularly comical that you ask if our disclosure is enabling after you just deleted the first tractographic image from the Wikipedia page. Enabling means that it works to produce the product and that is a diffusion anisotropy based, selective progressive tractographic image. You think you can just erase this, thus "disabling" our patent filed 17 years ago, and tell us all to go away because you are angry and uneasy to discover that your basic assumption - your "creation myth" is utterly incorrect. Afiller (talk) 16:17, 22 July 2009 (UTC)

Basic Truth - Going back in and editing my discussion comments above is cowardly masked vandal stuff - just for your information.Afiller (talk) 16:32, 22 July 2009 (UTC)

Afiller-Those companies can and will be willing to say whatever that pleases you or the examiners as long as they don't have to pay you a dime. Of course, their methods are more sophisticated because their methods are not based on DVI. BasicTruth (talk) 16:56, 22 July 2009 (UTC)

Basic Truth - We have been paid more than a dime by manufacturers, I can assure you. Your insistence on attributing our tensor work to Basser is totally unsupportable. That July 31, 1992 filing was written in Seattle in early July, submitted to the Department of Defense for security clearance for overseas filing, then transmitted to the UK for additional figures and signatures leading up to the July 31 filing date. All that is BEFORE any publication by Basser. I had been using standard tensor math for decades. Taking six inputs to calculate the orientation of the principal axis of an ellipsoid tensor was something I was taught in grad school at Harvard 10 years earlier - that is the point of the Tauxe reference. The idea that the historical first success is then the method that we must continue to use without change for next 17 years is a rather foolish understanding of how progress in technology works. This is a history section not a current methods section. The purpose is to show how the methods developed historically. At the same we document the progressive steps and critical major advances that set the stage for further development. The fact that CT scanning follows fire starting with flint by thousands of years doesn't mean that we use fire starting with flint in a CT scanner.Afiller (talk) 17:42, 22 July 2009 (UTC)

Afiller---The “operational definition” of scientific merit of one’s scientific works is the number of citations. Your works, by now a passé, did not get many citations (a mere 82 from ISI [July 22, 2009]) after all these years and even with the initial, perhaps undeserved, media attention. Based on your number of articles, it seems to me that you have difficulty working with others or getting things done. Oh.., the vector formulation did not even have your name, (Richards TL, Heide AC, Tsuruda JS, Alvord EC: Vector analysis of diffusion images in experimental allergic encephalomyelitis. Presented at Society for Magnetic Resonance in Medicine, Berlin, SMRM Proceedings 11:412, 1992 (abstr).). Hmm…, I wonder why. By the way, the filling date was after the 1992 ISMRM at which the abstract by Basser et al. was presented. It is a pity you have to resorted Hounsfield's stuff to justify your incompetence.BasicTruth (talk) 17:51, 22 July 2009 (UTC)

Basic Truth- The SMRM abstracts were published in August - after the application was filed - so your suggestion that I copied from Basser is not supported. I cover the issue of priorities, etc in the review article. We have an inventor group that each contributed and that is clear from the filing sequence and the review. In general, we don't measure inventions by citations. There are very few citations of Hounsfield's CT patents so by your standard CT is worthless. Priority is priority not popularity. Basser gives out grants at NIH so everybody loves to be nice and cite him. That's great. Who cites Isidor Rabi anymore - does that mean that magnetic resonance is "passé"? Not to compare myself - but citations is not the only measure of invention and merit. This is a history section and I have accurately told the history. I'm sorry that is upsetting to you. Fortunately, as you anonymously hand out insults, you can't change the past. Interesting process - but you have not invented the method of doing history by bullying, insults and telling everyone around you that only you hold the monopoly of truth - even if differs from the facts. Im sure you can see that we covered the substance and these exchanges add no light to the issue.Afiller (talk) 20:16, 22 July 2009 (UTC)

Afiller-Ask yourself whether the public has benefited from your studies---neuroticgram, oops, I meant that neurogram of yours. It is a pity indeed---what a waste of air time and public money. Creating preceeding and wiki now and after all these years of cowardly silence seem too frivolous. I am wasting too much time already, adios AFiller, I am out of here. May you find peace and quiet. Hope you don't get carpal tunnel syndrome with all the frivolous writings. BasicTruth (talk) 21:29, 22 July 2009 (UTC)

Basic Truth - Fortunately, tens of thousands of patients have been relieved of severe pain syndromes after identification of nerve injuries and entrapments by neurography. Neurography plays a significant clinical role around the world. I guess I can't see why that was a waste.Afiller (talk) 23:16, 22 July 2009 (UTC)

Afiller and Basic Truth (if you are still there)--- I have been reading up your stories and arguments. If I may, I would like to provide the facts I found. I don't agree with Basic Truth's aggressive approach but I also don't find Afiller's argument convincing either.

First, Afiller, your 1992 patent did not mention diffusion tensor and did not show how to computing the diffusion tensor. As Basic Truth said, "reason backward" is an important here and I did not see your 1992 patent or the series of papers published by you and your colleagues from 1992-1994 contained any technique for doing the "inversion". I have been in the field for quite a while. I have to admit I did not quite get how the diffusion tensor is computed or estimated until a later paper by Basser and Pierpaoli (A simplified method to measure the diffusion tensor from seven MR images, 1998).

Second, Afiller, I agree with you that the scientific or commercial merit of an invention should not be judged the number of citations. Of course, Hounsfield is well-known and so is Mansfield. Mansfield's initial papers in MRI do not get cited a lot, and I believe, they are too well-known to warrant a citation. The same situation applies to Lauterbur. Afiller, your situation is different because your works are not well-known or at least we are not aware of them until you made them known through this Wiki. Why it took you so long to make some noises?

Basic Truth, Please only put appropriate materials here. Cool down. Best, CogitoErgoSum101 (talk) 16:12, 23 July 2009 (UTC)

CogitoErgoSum101 - Thanks. I go through it in technical detail with all references in the review article - so I would suggest you look through that because I can't put 151 references in this reply. However, the issue in technology development here is the idea = widely held at the ISMRM - that Peter Basser "from the mountain top" gave us DTI in 1994. From our point of view, writing two years before that and unaware of his work as he was still publishing about strain guages at the time - we saw the problem of how to generate a rotationally invariant tractographic image. As the review shows we were interested in tractography and began to deploy diffusion. The tractographic vector image shows that the major conceptual and technical advances from the separate planar images are not in the tensor calculation per se. I had extensive experience with the six input tensor solution so we thought that was all well known in many fields. That is why we say in the patent that it is well known how to do the tensor solution - because it was well known and I cite a number of references on this. Also - as you will see from the reference in the review, we set up a company - NeuroGrafix to develop the technology as best we could. Our chief scientific officer, Jay Tsuruda was heavily involved in research on eddy currents and other issues at University of Utah - so we did everything we could. I also made the correct decision - I think history shows - that the most immediate clinical applications in 1992 and 1993 would be in peripheral nerve. As I say above, we have had an enormous impact on thousands of lives - beginning from a time when clinical DTI was a dream because the clinical scanners just didn't have the gradient strength. I don't deny the value of all the wonderful work by Denis Le Bihan, the ingenious mathematical treatments by Peter Basser and the incisive work by Susumu Mori in understanding how to carve up a complex problem and force it into a nice executable solution. I don't have a problem with any of that. I didn't really realize we weren't being cited until I began looking into Peter Basser's work in more detail. I say more than enough about this in the review. You will learn that Michael Moseley, Graeme Bydder, John Griffiths and many other in MR are very well aware of what we did then and what I and my co-inventors have worked on. Jay Tsuruda was with Moseley in 1990 and continued in this field. Franklyn Howe is an Oxford trained MR physicist who was in John Griffiths lab - his focus has been cancer. He showed how to minimize cross products when we added the mulitple diffusion gradients. Todd RIchards showed a way of solving many of the fundamental issues. I thought we had done our job and history shows that we did. We solved some critical problems, Moseley made everyone aware of what we did. The plenary session I reference had just three speakers. Moseley spoke about diffusion, Le Bihan was the moderator - everyone there saw the tractographic image and heard how it was done - the image that Basic Truth thinks no one should see. I can't compete with 1,000 well funded MR scientists for sheer production. I try to innovate in various fields and do what I can to try to be sure that my invention "grows up." My father invented much of the technology behind shock wave shaping and beaming in water. This led to lithotripsy. He continued inputs into this. However, he explained to me that an invention is like your own child - no one cares like you do about its success and it's your responsibility to give it what it needs to succeed. I think we did that with this technology. As you can see the ellipsoid tensor formalism had its role but we thought it would not work well because in peripheral nerve branching is important. I have a paper coming out soon that explains the anti-symmetric dyadic tensor model I had in mind. In general, you can't patent math. However, we feel and the patent office agrees that given what we said in the patent and what was known at the time, it no longer required "inventive input" to see how to do tractrography - but many inventive improvements can be added. Mendel did very little to promote genetics but there are reasons why he is very important. I was surprised by Basic Truth's response - I think it has to do with undermining his "creation myth". He argues that our solutions - showing how to accomplish a rotationally invariant calculated image, showing how to tract trace, pointing out that inputs from an infinite number of gradient directions can be fed into a tensor math calculation. The tract tracing should be selective and progressive. We explain that you choose a starting point and target and then run the algorithm to connect the related voxels he argues that all that is some kind of offense rather than inventive contribution. It wasn't known before, it is all required, it is not OK to say that others who heard about it forgot and reinvented it.

There never was a history section on this article. I added one. Nothing is inaccurate. The man who controls the grants in this field - Peter Basser - will certainly be pleased if you conclude that all of this relevant history should be removed not cited and not discussed. They introduced known diffusion tensor analysis of pork loin related known muscle fiber anisotropy. I do think this had a big influence. You could write that up in more detail and remove any reference to our well publicized contributions. That is one reason why patents exist and why scientists bother to file them - there is no good way to erase them if you want to and priority gets adjudicated. I'll try to stay settle down and answer the questions - but it will help if you read through the review (now available on Nature precedings) and keep in mind that your own received view _- the creation myth -might be incorrect.Afiller (talk) 17:43, 23 July 2009 (UTC)

Afiller – First, could you please provide the name of the Counsel for Philips who allegedly "clearly accepts" the claim that your group invented DTI and tractography?

Second, what is CLEAR to me is that you don't understand the physics and mathematics behind DTI or the unique information it produces. Before the diffusion tensor MR framework was introduced by Basser et al. there were many unsuccessful attempts to characterize 3-D displacement profiles using DW data. The work of Richards et al. on Diffusion Vector Imaging (SMRM, 1992) is an example of such a failed attempt that was presented at the same SMRM meeting as Basser et al.’s two seminal abstracts on Diffusion Tensor NMR. (As an aside, you are not even listed as a co-author on Richards et al.’s SMRM abstract that introduces and describes the same framework you claimed to have invented!). Your claim that "Taking six inputs to calculate the orientation of the principal axis of an ellipsoid tensor was something I was taught in grad school", makes me wonder whether you failed the course, since your proposed approach only uses THREE directions to try to provide that information. The framework proposed by Richards et al. cannot generally provide the principal axes of the diffusion ellipsoid or any other quantity obtain from the diffusion tensor, expect possibly its Trace. (Do you know what that is?) Please, for the sake of those of us who did not fail linear algebra and took it in high school, refrain from confusing a vector with three components and a second rank tensor. Cheers Natasha09 (talk) 19:38, 23 July 2009 (UTC)

Natasha09 - Thanks. The vector method is not the only method we covered in the patent. It has a historical role. We also pointed out how to get the data for the tensor - the math to process it was well known. The vector example solves a variety of important problems that enable imaging. I can only say it so many times - tensors - including six components were well understood at the time. We explained how to collect multiple gradient orientations because we understood that you need the additional axes for the tensor and we state this. All the invective won't make it disappear. People are accustomed to thinking of the origin of DTI as if Basser discovered how to compute a tensor. He didn't discover that and - that was not the only problem to solve to accomplish tractography. I am sorry this makes you angry. However, as I pointed out at the outset, the best thing to do is to delete everything about what we did and everything written and published - then it won't trouble you. You have to actually read the patent and the July 31 filing before you get too excited about what you think is not in them. For instance - the abstract is one page, but the patent is 42 pages. The abstract covers vectors, the patent covers vectors and tensors.Afiller (talk) 21:09, 23 July 2009 (UTC)

Natasha09 - Your other question was about the patent attorney. Numerous patent examiners in the US, Japan and elsewhere have cited the anisotropic imaging patent - along with Susumu Mori's patent - as the basis for DTI in the patent literature. This is why the 5,560,360 patent (diffusion anisotropy imaging) has been cited as critical prior art for the vast majority of all 60 or 70 patents in this field. The attorney for Philips - as documented in the published and publicly available file history for US 6,724,190 didn't have much choice because that fact was never in doubt, they were just arguing about interpolation methods. Many scientists have a very difficult time understanding patents and technology advancement because the process is so different from "normal science." The science paradigm is that we all agree what needs to be done, we all agree how to do it, we all agree on what lab and individual is best equipped to do it, and then that person in that lab used the equipment we all have and gets the result we all expected that person to get. Inventions have to bridge gaps that leave good scientists puzzled. Try to ask what it was that all those excellent MRI scientists in 1991 needed to know to get from their three separate images to an operational tractographic process. Everyone has thought that all that was needed was for Peter Basser to tell the MRI scientists how to compute a tensor like everyone in other areas of science needed to do. I don't think so. The Richards experiment is a vector solution (just like Q-Ball is a vector method) - but it does not make it a useless advance - you have to consider the status at the time. It is not the same paper that Basser and LeBihan wrote - it solves a different set of important problems in getting from pixel data imaging to tractography. If you know tractography well, you will know that we first use a method to get the orientation of the principle axis and then we throw away all the other axis information. Once we have the orientation we have a separate task to accomplish the tract tracing. The usage in the patent is to show that we can use vector or tensor methods to get the principle axis direction and magnitude, and then we deploy a tractographic algorithm. Do you think there were any tractographic algorithms before the Richards abstract? How clever was it to come up with tractography if you had the Richards abstract and the Filler patent in hand with the basis of the solution already worked through. All you can do is to come up with more an better ways of accomplishing the tractography but you can no longer invent that category of basic process. I have a difficult time seeing what problem the Basser paper solved other than showing other scientists how to plug the diffusion measures into the Mathematica tensor calculation software.Afiller (talk) 00:41, 24 July 2009 (UTC)

Natasha09 - One more thing - I believe that when people express opinions about Wikipedia content that it can be anonymous. Name calling and insults are only OK if you abandon anonymity and say who you are. I don't mind it, but I'm sure you want to stay anonymous because you know it makes you look silly and rude. I'm all in favor of irreverence, humor and irritating sarcasm. However if you rely on calling someone a dummy then it suggests you don't have a specific well thought and researched point to be made.Afiller (talk) 01:04, 24 July 2009 (UTC)

Ergo & Natasha - In trying to understand the significance in early 1993 you have to ask yourself why would Professor Moseley take up 2/3 of the plenary session on diffusion at a major meeting talking about what our group had accomplished if it was such obviously poor quality work?Afiller (talk) 04:35, 24 July 2009 (UTC)

CogitoErgoSum101 - I'm afraid that this is another example of a reader being sufficiently upset as to miss what is clearly stated. You say above that we did not mention the tensor in the 1992 filing. I have provided a link previously - that still works fine - that lets you download the July 31, 1992 filing. OCR has been run. You can get a copy from the UK patent office or from Delphion if you don't believe it. It is GB9216383A. On Page 21 of that filing, here is sufficient text:

"The use of vector analysis algorithms of this sort, or involving the treatment or coordinate transformation of MR diffusional anisotropy data with tensors of various rank can improve the generality and flexibility of neurographic imaging. The example described above demonstrates that by the application of tensor and/or vector analysis methods such as algorithms similar to those developed for the evaluation of e.g, magnetic, thermal, or structural anisotropy data, it is possible to greatly improve the flexibility and generality of image techniques for neurological diagnosis."

We go on to say:

"If the major axis of diffusional anisotropy is not known, a plurality of images are acquired with different orientations of the diffusion sensitizing gradients. The image data can then be processed to give a parameter associated with each pixel (or voxel in a 3D data set) which is a measure of the diffusional anisotropy at that point reflecting both magnitude and direction."

As with anything else, I think it is important to read through the entire document - the US 5,560,360 patent before going online with authoritative sounding statements about what is missing from it. Remember - if you read the Tauxe reference and consider that this was work I lived with day in day out ten years earlier as a grad student at Harvard (the six axis tensor solution) - when I say "algorithms are similar to those developed for the evaluation of e.g. magnetic data" it is fully enabling. Tauxe - to save you time - fully describes the six axis tensor solution. The paper is commenting on Jelinek's older paper that says that more axes are needed - typically 12 axes - all of this is in the review.

In US 5,560,360 we say:

"gradient coils oriented in three planes can be simultaneously activated In various combinations to achieve the effect of an infinite variety of differently oriented gradients." So there is no point trying to say we had no understanding of more than three axes for calculating the orientation of the tensor. This differs from earlier work in that it points out the multiple gradients, comments on using tensor analysis to get the orientations and then shows the additional math needed to take the principle direction and magnitude info forward into a tractogram and we showed what one looked like. Hopefully this makes it more clear why the patent examiners believe we invented DTI and tractography. You are not going to get a 2006 quality solution in 1992, but this was all an enormous advance over what was done in other groups (showing three separated images). According to Basser's interview (see the review for the reference) he became dejected when Le Bihan pointed out to him that although it should be a tensor, no one knew how to measure the tensor. Further Basser states that as late as 1994, tractography seemed like science fiction to him. Make your own judgment on priority here, but I stand by my position and a calm thorough reading will support the position I have taken and the history as stated.Afiller (talk) 05:55, 24 July 2009 (UTC)

Natasha09 - Since you are so sure of your geometry, it is worth pointing out that you are quite incorrect - you can indeed get the Euler angles from the three input vector solution. I explain the issue in detail in a paper coming out in Neurosurgery in a few months - (It is my other article on Nature precedings and cited in the Diffusion MRI Wikipedia article). The real problem arises because of the bidirectionality of diffusion and the fact that an intensity measured by a gradient acquisition therefore cannot distinguish flow towards and away from the point of view. The result is that Euler angle calculations in diffusion MRI are often incorrect - this is why it is inefficient. If you understand the nuts and bolts of tractography algorithms, you will know that if you ran a total tractogram - even with the best tensor data - you would get a mass of correct and incorrect tracts of various lengths. We have to choose a seed and destination in many of the algorithms, then reject all the solutions that do not result in a tract that runs from the seed to the target. It is messy, but it works. The same process worked with Euler angles from the Richards vector analysis. That is why it results in a tractogram. If you were right when you say that vectors can't produce a Euler triplet, then the Richards tractogram - and a vast mass of other working systems in astronomy and computer graphics - wouldn't actually work. It's all moot anyhow because you failed to read the patent before stating that it didn't cover tensors. You might have other questions or objections, but - like Basic Truth - you are just upset because your creation myth is collapsing and you want to strike out to protect it.Afiller (talk) 07:42, 24 July 2009 (UTC)

Please: It's "an Euler triplet", not "a Euler triplet". The name is pronounced like "oiler". Michael Hardy (talk) 00:32, 22 May 2014 (UTC)

Natasha09 - To allay your concerns on the Trace and to help generate a better Wikipedia article - I have added a section explaining trace, radial diffusivity, and other additional indices and further explaining the derivation of fractional anisotropy. Afiller (talk) 07:03, 26 July 2009 (UTC)

Cogito, Natasha & Basic - I have added a section providing a basic level explanation of the diffusion tensor and how it relates to the relevant ellipsoid math, to the matrix and vector processing steps for deriving the eigenvectors and eigenvalues and to the evaluation of the principal axis for tractography.Afiller (talk) 06:30, 5 August 2009 (UTC)

Some comments above imply that Peter Basser administers grants for NIH. He in an intramural scientist. The extramural program at NIH administers grants. Intramural scientists have no role in administering grants.Rothbrad (talk) 01:14, 8 May 2010 (UTC)

Method of FA calculation

Susumu Mori in his book gives 1/2 as the factor (Introduction to Diffusion Tensor Imaging, p.70, equation 7.1). Basser et al provide 3/2 originally. Apparently both methods are used in the literature. This is not a strict mathematical principal rather, it is an empirical method for scaling the FA numbers. By using a factor of the square root of 1/2 (0.7) we bring the fraction closer to 1. By using the square root of 3/2 (1.22) we increase the number so it is more often greater than 1.Afiller (talk) —Preceding undated comment added 23:42, 22 August 2009 (UTC).

In 2007, Basser also uses 1/2 as the factor - see Chang, Koay, Pierpaoli, and Basser, MagResMed 57:141-149, (2007). Since Mori says 1/2 and Basser now says 1/2, it is fair to say that 1/2 is correct. It also makes more sense when you consider the intention to try to describe the extent of anisotropy as a fraction of the unit vector. Afiller (talk) —Preceding undated comment added 08:09, 23 August 2009 (UTC).

"Howto"

A discussion on this article's "how to" style (or lack thereof) is being carried at Wikipedia_talk:WikiProject_Mathematics#Really_strange_fork. Please reply there to keep it all in one place. Pcap ping 12:31, 3 September 2009 (UTC)

I've done some copy edits in the article, down to the final, mathematical section. In the light of what was said over there, I don't want to be heavy-handed in changing that all around. But some work is clearly needed in organising that section to read better. For example, the history para including Fick seems to be floating currently. Charles Matthews (talk) 09:21, 6 September 2009 (UTC)

The final section seemed to need subsections to sort the material out. I now think the "Mathematics of ellipsoids" subsection could do with further copy edits, for tone, but generally the article is not so bad (work needed in a few places). Charles Matthews (talk) 09:32, 6 September 2009 (UTC)

Thanks to Charles Matthews. That is some helpful copyediting. I'm an Oxford University Press author so naturally suspicious of editing coming from Cambridge. There is a great deal of interest in this area and it will grow. Hopefully your time and effort in this will lead to a little bit of permanence for the text. It gets visited by hundreds of experts as well as those just trying to get their feet on the ground in this area. I also hope the mathematicians will be soothed a bit!(Afiller (talk) 04:38, 7 September 2009 (UTC))

dipy?

Why is "dipy" listed under 'see also'? It is not notable relative to other packages, and as such the presence here could be construed as promotion. If packages are to be referenced: DTIStudio, MedInria, Slicer, and FMRIB-DT all have dti-related citation counts in the 300-600+ range per Google scholar. (Camino falls in that range). "dipy" has none. —Preceding unsigned comment added by 65.96.161.120 (talk) 06:11, 26 April 2011 (UTC)

Too technical/length

I happened across the article from a random search and it strikes me as a bit technical for the average reader to comfortably understand (that is, to not need to learn about its full scientific background before being able to understand/grasp the article). Is it just me or are there ways to rewrite things to make them somewhat more...layperson-minded? (See WP:MTAU) I hesitate to use the {technical} template since I'm not exactly sure as to what specifically could be done to simplify things but feel it should be brought up.

Also, the article is a bit long and seems to attempt to cover everything related to dMRI and similar technologies. One way to cut things might be to give an intro to 'Diffusion tensor imaging' but split that section off into its own article. (See WP:SPLIT and WP:LENGTH) Coinmanj (talk) 08:41, 31 December 2012 (UTC)

Gone?

Filler? Have you gone now? Can we write an apolitical article on the topic of Diffusion Weighted Magnetic Resonance Imaging and it's application to clinical and basic brain science by hundreds of thousands of doctors and researchers yet?

because we're all waiting for you to stop grandstanding so we don't waste our valuable time with edit wars fuelled by how much Peter Basser (or whomever) pissed you off 20 years ago or whatever other irrelevance you are pushing.

Realize this is a public service by volunteers seeking to disseminate useful information. In this case, the necessary volunteers are as busy as you. Aren't they?

Just let us know. This article is very poor right now. People aren't editing it and it's mainly because of you.

Fine you invented it, you're very clever and the universe owes you billions in patent infringement. can we move on from the irrelevant past? Duracell (talk) 22:01, 3 August 2013 (UTC)

First two sections are completely inaccessable to the lay reader

Wikipedia is for the lay person so the first two sections really need some explaining - not great as they are right at the beginning. Perhaps we should put them later on or at least explain what is going on! 159.92.170.167 (talk) 12:56, 16 January 2014 (UTC)

Basic information

It's great that I now know how to calculate anisotropy values by hand if and when I build an MRI in my living room, but it's too bad that in ~10,000 words of technical description and equations, this article lacks the basic detail I came here for - whether fractional anisotropy is directly or inversely related to white matter integrity. The article does not seem to be designed for people who are not already experts on the topic. 50.165.204.29 (talk) 20:43, 19 January 2014 (UTC)

Regrettably not very simple! Diffusion metrics combine all the information in a voxel, which will have many fibres and possibly multiple tracts in it. In the case of crossing fibres (very common, a quick google gets this http://www.ncbi.nlm.nih.gov/pubmed/22611035 65-90%, but, as they say in the abstract, this is much higher than previously reported, 30% was a more commonly quoted figure) the overall FA is lower than the FA of a single fibre population, so losing one set of fibres and not another could cause FA to increase depending on the situation. Demyelination or loss of fibres would allow freer transverse diffusion, causing decrease in FA. Other possible influences are inflammation, see discussion in http://brain.oxfordjournals.org/content/early/2013/03/22/brain.awt065.full relating to Alzheimer's (link should be free access). The basic information on calculation is probably quite useful for researchers starting out in this area. 193.62.68.5 (talk) 17:10, 12 September 2014 (UTC)

Typo in ADC

What does "attenuated that diffusion is fast" mean? Also, it might help to explain what x, y, and z are in the ADC equation.

Male/female differences in Pennsylvania PNAS paper

"In 2013, researchers from the University of Pennsylvania reported in the Proceedings of the National Academy of Sciences a study to map differences in male and female brains.[59] The report demonstrated that in male brains, intralobe neural connections dominate, while in female brains interlobe connections dominate. It was also demonstrated that this distinction is acquired rather that inborn.[60][61]" - This was the press release version but not in fact what was found (connections were demonstrated to be stronger according to gender in this direction at a level that was statistically significant but not large compared to population variation, 'connections dominate' is not the same), even without pointing out some flaws in the analysis. 193.62.68.5 (talk) 16:06, 12 September 2014 (UTC) a.k.a. Imalone (talk) 09:51, 13 September 2014 (UTC)

Sorry, could you just link to the paper? PMID is best. -- CFCF 🍌 (email) 17:55, 12 September 2014 (UTC)

[1] (currently cited [59]) does not give estimates of the difference, only p-values and T statistics for the group difference, which if a difference exists can be made arbitrarily small and high respectively, they have 949 subjects. You can estimate the effect size from their numbers though, which looks less impressive [2] the largest methodological problem was failure to control for head size, which might explain much of their effect, [3] (don't know much about science media centre as a source, but Heidi Johansen-Berg is a well regarded author in the field). There's no comparison between different tracts either, so the repeated statement that certain connections dominate isn't supported. The citation is also repeated on Neuroscience_of_sex_differences, but since I'm not a regular editor here I've no wish to blunder in and change it. Imalone (talk) 09:51, 13 September 2014 (UTC) ...Also a reasonable summary in this frontiers in human neuroscience article Imalone (talk) 10:10, 13 September 2014 (UTC)

this is a WP:PRIMARY source and fails WP:MEDRS - shouldn't be used anywhere in WP. Also the article is not about DTI per se, but rather a finding generating using DTI - there are a few hundred such papers and it is pure WP:OR to select this one out of all them to discuss here (if we decided to go down that path, and I don't think we should) But if we did have a section on "Uses" or the like, it should be sourced from a review and should provide a broad overview of the various uses that are being found for DTI. Jytdog (talk) 12:06, 13 September 2014 (UTC)

Just to be clear, I'm not suggesting citing this paper, I'm pointing out it is already cited and is somewhat contentious. The newspaper reports referenced also exaggerate the actual findings of the paper. There could certainly be scope for a round up of review articles of applications and findings, but since I'm writing this on a phone probably won't stay today :) Imalone (talk) 11:30, 14 September 2014 (UTC)