idea: MRI Clearing House

97 views
Skip to first unread message

Yaroslav Halchenko

unread,
Jun 28, 2021, 9:16:29 AM6/28/21
to niQC
Hi All,

I have been thinking on what could be a project/effort which would take advantage of the majority (if not all) the efforts of folks in the group, with some clearly defined target/goal.  What do you think about:

MRI Clearing House: an archive and analytics platform aggregating MRI Phantom QC data from participating imaging centers.

Some principles to guide:
- exchange:
  - of data (QC, experimental WiP sequences)
  - QC pipelines
- automation:
  - data submissions/analysis should be automated as much as possible (may be even by providing a PACS server to receive data)
  - those various QC pipelines could be ran for each center's data
  - reduction of necessary work needed at individual centers
- awareness ("early warning" notification system):
  - changes in data QC etc detected at the (other centers) scanner software version changes
  - variability across centers (and thus possibly across/within studies etc)
  - providing variables worth including in "human studies" as potential variance explaining variables (see https://f1000research.com/articles/9-1131), thus contributing to reducing "noise" and increasing power (without crying out for "more samples")
   - side effect (untested hypothesis): one center's 20 subjects might be another's 40.  Providing centralized public dashboard of centers' QC could be useful to provide support for smaller sample studies or demand more samples in "noisy centers".

With a sufficient adoption by the community, we might even make manufacturer's get interested to monitor/contribute to the system. 

Sounds ambitious?  Good! 
I think we could start "little" by first establishing some data collection/archival etc. I could propose some technologies and provide some infrastructure. 

Sounds familiar? Even better!
Please share pointers to already existing initiatives/solutions I could join or contribute to.

Interested? Awesome.
Let's get the ball rolling.  Will add this topic to Pradeep's "webinar series" agenda.

Cheers,

Alexandre Franco

unread,
Jun 28, 2021, 11:13:02 AM6/28/21
to niQC
Hi Yarik, 

Sounds like a great idea. Let's schedule this discussion to get the ball running. 

Best, 
Alex

Aina Puce

unread,
Jun 28, 2021, 11:21:06 AM6/28/21
to Alexandre Franco, niQC
Yarik,
A discussion would be welcome!

[I am not totally sure the reach of this list, but some input from those with legal experience would also be helpful, or from ppl who had links with certification of medical hardware/software [e.g. the International Electrotechnical Commission in Europe, FDA in the USA & Standards Australia...]
There are potential legal issues here for usage & I want to make sure that we do not overlook those & hamper progress for this wonderful idea...
Aina

--
You received this message because you are subscribed to the Google Groups "niQC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to niqc+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/niqc/0ce1a55b-81c6-45a9-baf0-ba974dba498en%40googlegroups.com.


--
Aina Puce, PhD . 

A button with "Hear my name" text for name playback in email signature 

Eleanor Cox Riggs Professor
Psychological & Brain Sciences
Programs in Neuroscience, Cognitive Science
Affiliate, Indiana University Network Science Institute
Indiana University, Bloomington IN 47405 USA
--
Chair, Organization for Human Brain Mapping
Associate Editor, Perspectives on Psychological Science
--
email: aina...@indiana.edu
Twitter: @aina_puce
tel (cell): +1 812 650-2213

Raamana, Pradeep Reddy

unread,
Jun 29, 2021, 9:23:27 AM6/29/21
to Aina Puce, Alexandre Franco, niQC

Hi Yarik, thanks for sharing this idea!

 

A phantom focused database is very interesting and worth discussing. Let’s schedule it for the QC webinar series - I know few others interested in this and please suggest other speakers/panelists/papers on this topic. I know of few efforts at few BIG imaging centers do some sort of monitoring (although with different interests and/or at different levels) – HCP pipelines/fBIRN, LONI QC and the BrainCODE platforms come to mind, and I am sure there is few others.

 

For the benefit of everyone, I’d like to remind we shared/linked to a nice phantom dataset from the Ontario Brain Institute here on niQC website. Please help add other resources to the niQC resources website.

Raamana, Pradeep Reddy

unread,
Jun 30, 2021, 7:33:15 PM6/30/21
to Aina Puce, Alexandre Franco, niQC
Also a related idea that may overcome any potential legal issues (that Aina cautioned us about) associated with such a centralized platform is decentralized processing and federated learning of useful features and models.  
From: Raamana, Pradeep Reddy <RAAM...@pitt.edu>
Sent: Tuesday, June 29, 2021 9:23:24 AM
To: Aina Puce <aina...@gmail.com>; Alexandre Franco <eng.f...@gmail.com>

Stephen Strother

unread,
Jun 30, 2021, 11:24:48 PM6/30/21
to ni...@googlegroups.com
Following up on Pradeep's comments, here is the preliminary publication based on the Ontario Brain Institutes fBIRN data set (https://www.sciencedirect.com/science/article/pii/S1053811921004742).

I believe a challenge with Yarik's idea, which I strongly support, is the lack of a affordable standardised phantom so the inter-scanner/site effects are not also phantom effects. As far as we could tell a few years ago the fBIRN phantom was no longer available from the original developers to buy for fMRI phantom scanning and none of the commercially available phantoms I am aware of (e.g., https://www.phantomlab.com) are particularly useful for the brain since they all are 20 cm in diameter. It is also relevant to note that the Phantomlab maintains a cloud based image analysis centre (https://www.phantomlab.com/smari-image-analysis) that is potentially aggregating data across scanners, but I know nothing about it other than that it exists.

At a minimum just finding out what most research centers are doing for regular scanner QC, particularly how and which primary parameters they are monitoring and how often would be a great start, i.e., a landscape survey.

Cheers, Stephen
To view this discussion on the web visit https://groups.google.com/d/msgid/niqc/DM6PR04MB6746A546CF6C5F0C1757EA80B8029%40DM6PR04MB6746.namprd04.prod.outlook.com.


-- 
----------------------------------------------------------------------------------------------------------------
To reduce e-mail overload follow the e-mail charter of Chris Anderson: http://emailcharter.org/.

Stephen Strother, PhD
Senior Scientist, Rotman Research Institute, Baycrest
Professor of Medical Biophysics, University of Toronto
E-mail: sstr...@research.baycrest.org
Tel. Office: 416-785-2500 x2956
Fax: 426-785-2862

Yaroslav Halchenko

unread,
Jul 1, 2021, 11:08:46 AM7/1/21
to niQC
A concern I have not had in mind to say frankly.  I think, unless we would somehow end up in a "certification" business, I think we should not be anyhow "approved" by some regulatory  body like those.  I would treat this project as a "car review" portal  as opposed to a "car manufacturer".  Overall, this project would be most similar to existing MRI sharing and processing portals and initiatives (openneuro etc), just collecting and operating on phantom data.

Yaroslav Halchenko

unread,
Jul 1, 2021, 11:27:50 AM7/1/21
to niQC
Fresh publication: NICE!  yet to digest fully.  It contributes to my argument of the importance of regular QC and keeping that data around!  I had reached out to some centers while looking for human + phantom data, and in a few cases the answer was "QC data was deleted X months ago since was not considered to be useful beyond its original QC purpose at that point in time".
Is QA data used in that paper available somewhere?  I think to get ball rolling we might start with establishing an aggregated collection of phantom data from interested to participate centers. Ours from DBIC is at https://datasets.datalad.org/?dir=/dbic/QA but I need to update it with freshier data.

Lack of standardized phantom: 100% a valid concern! My thinking ATM:  (meta)data for each "phantom subject" should have sufficient information about that phantom "class", "age", etc, so we could at least "group" data from "similar" phantoms. I would expect many centers to use a similar-ish phantoms (e.g.  agar ball Phantom like we do) and thus could already be somewhat comparable.  Eventually we could even extend the effort with "traveling phantom" run, where some phantom(s) would go   through participating centers and possibly those scans would allow to establish "transformation"/"alignment" between used in house phantoms of the same class and this "traveling phantom".   In future we might even encourage harmonization of phantoms (and sequences) used for QC in centers toward some common recommended set.  So -- altogether -- yes a concern, but I believe we can address it "one step at a time" ;)

Even without phantom correspondence, there are additional target uses for such data/metrics sharing:
- studies could refer to their phantom QC data for the duration of any particular study, so intrigued researchers could get a better glimpse at the health of the scanner through the duration of any particular study (without cross-referencing to any other phantom)
- any particular phantom data QC metrics abnormality which could be attributed to e.g. change in software version, humidity, temperature (if collected/provided) could be an alert to other centers
- QC metric model derived from one center might still be applicable to be transferred for validation to data from another center with data on the phantom of the same "class" and comparable scanning parameters.  

https://www.phantomlab.com : good to know.  We might want to figure out how to make friends with them at some point for the mutual benefit, as opposed from them being hurt, by our effort. E.g. we could integrate with their service to offload data from accepted by them phantoms to their platform as well for participants who are subscribed to their service. Their "proprietary" (?) metrics/artifacts detection comparison to "ours" might be shown to be "superior" in terms of detection power etc (we don't know) thus giving them a good exposure. Etc. 

Yaroslav Halchenko

unread,
Jul 1, 2021, 11:34:19 AM7/1/21
to niQC
"Apply for access on BrainCODE." .

That is what I wonder -- how many centers would be willing to openly share their phantom data!?  IMHO it is not a subject to any "human subject protection" regulations, and overall the resistance toward open sharing of the phantom data would be "we are afraid to disclose that our scanner actually sucks".  Any other valid concern I am brushing off?

Hypothetically speaking: (might be a nice study actually!!!) If I have access to phantom QA data, I might be able to cross-reference human data by analysis of e.g. outside of the head "noise" and thus identify likely ranges of dates when that human was scanned (which might be a concern for HIPAA's "patient visit" being sensitive data). But since QC is not done daily AFAIK, even with exact QC scan date identification it would still be week or two window I guess. 

Stephen Strother

unread,
Jul 1, 2021, 9:48:38 PM7/1/21
to ni...@googlegroups.com
Another study that I am aware of that is directly relevant to this discussion and is based on a large multi-site set of phantom data is https://www.sciencedirect.com/science/article/abs/pii/S1053811915006199?via%3Dihub. In particular see Fig. 2.

Raamana, Pradeep Reddy

unread,
Jul 3, 2021, 3:47:13 PM7/3/21
to Stephen Strother, ni...@googlegroups.com

As I try to follow this discussion, I am wondering, after all these years of MRI scanning across the world, why don’t we yet have a standardized phantom? Is the explanation the obvious one of large multi-site studies and the associated challenges never properly receiving sufficient consideration?

 

I very much like Stephen’s suggestion of phantom-focused community survey: “At a minimum just finding out what most research centers are doing for regular scanner QC, particularly how and which primary parameters they are monitoring and how often would be a great start, i.e., a landscape survey.”

 

We did a very broad niQC survey before, but I feel a much more targeted survey followed by a paper “review of phantom-based QA” would be very useful and straightforward projects, that can aid in the direction of the larger idea Yarik put forth (centralized database with some monitoring and archive features).

 

Also a gentle reminder, we have a Zotero library collecting the relevant literature to niQC here: https://www.zotero.org/groups/2221093/niqc?

 

Please feel free to add to it, if you are a Zotero user (which I encourage you to be as it is free, cross-platform and open source). If you know of even a better way, let us know 😊.

Jean Chen

unread,
Jul 5, 2021, 2:15:48 PM7/5/21
to Raamana, Pradeep Reddy, Stephen Strother, ni...@googlegroups.com
Coming late into this discussion --- congrats Stephen on your recent phantom QA paper! 

These are some of the commonly used phantoms I've encountered: 
-- The ACR phantom is common enough that it may be a close surrogate to a "standardized phantom". It can be useful for things like SNR, ghosting, gross distortions, blurring, slice gap, etc, but is not suited for geometric distortion correction and fMRI. 
- The Lego phantom has been purchased at many Canadian sites, and its enhanced spatial information is more amenable for estimation and correction of fine geometric distortions, but this phantom is not made for fMRI
- The NIST phantom is for QC of quantitative imaging, and is not necessary for sites that do not do quantitative PD, T2 or T1 analyses. 
- The fBIRN phantom was developed for fMRI (allows for assessment of tSNR, so can also be used for other EPI-based acquisitions), as it is spherical shaped and contains tissue-mimicking gel. While these are advantages compared to the NIST or Lego phantoms, it does not have spatial structure and is non-quantitative. 
- ASL and DTI phantoms: would need to cater to the high gradient requirements, currently uncommon/unavailable. 

As you can see, there might be a need to first define "QC". Anatomical phantoms are QC are more relevant for clinical sites, which mostly run T1, T2 and FLAIR, but fMRI and DTI QC are increasingly becoming necessary, and while it may be unfeasible for each QC scans to involve multiple phantoms, I think we as a community know enough to realize that the necessary phantom(s) and QC protocol(s) depends on the type of scan. The paper that Pradeep proposed might want to address this realization --- or perhaps each large-scale study should designate its own set of QC phantoms like ONDRI is doing, to be scanned separately from the site QC session? 

Jean

--
Jean Chen, Ph.D.
Senior Scientist, Rotman Research Institute, Baycrest Centre for Geriatric Care
Canada Research Chair in Neuroimaging of Aging
Associate Professor, Medical Biophysics, University of Toronto
Full Member, Institute of Medical Sciences, University of Toronto





Simon Duchesne

unread,
Jul 5, 2021, 3:58:40 PM7/5/21
to Jean Chen, Raamana, Pradeep Reddy, Stephen Strother, ni...@googlegroups.com

Late as well, just to state that (sorry Yarik… I know it takes forever) at some point we will release the LEGO phantoms that have been collected haphazardly at the 26 sites of the CCNA. Haphazard, since not all sites are diligent. But, it also comes with the CCNA dataset. Hence, it would be great to have techniques to do phantom-less distortion QC straight from the images themselves. One can imagine learning the joint distribution of phantom and image distortions, for example.

 

Happy hunting

Simon

 

 

From: <ni...@googlegroups.com> on behalf of Jean Chen <jc...@research.baycrest.org>
Date: Monday, July 5, 2021 at 14:15
To: "Raamana, Pradeep Reddy" <RAAM...@pitt.edu>
Cc: Stephen Strother <sstr...@research.baycrest.org>, "ni...@googlegroups.com" <ni...@googlegroups.com>
Subject: Re: [niQC] Re: idea: MRI Clearing House

 

[Externe UL*]

Aina Puce, PhD . 

Image removed by sender. A button with "Hear my name" text for name playback in email signature 

--

You received this message because you are subscribed to the Google Groups "niQC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to niqc+uns...@googlegroups.com.

*ATTENTION : L’émetteur de ce courriel est externe à l’Université Laval.

Évitez de cliquer sur un hyperlien, d’ouvrir une pièce jointe ou de transmettre des informations si vous ne connaissez pas l’expéditeur du courriel. En cas de doute, contactez l’équipe de soutien informatique de votre unité ou hamec...@ulaval.ca.



Raamana, Pradeep Reddy

unread,
Jul 5, 2021, 6:04:48 PM7/5/21
to Simon Duchesne, Jean Chen, Stephen Strother, ni...@googlegroups.com

Thanks a lot Jean and Simon for the useful insights and pointers. From where I stand it appears to me we have enough experts and existing work for a “short afternoon workshop” of sort (few short talks, followed by an open discussion) focused entirely on phantoms as they are used in neuroimaging QA and QC. That would help us realize a clear outline for the phantom-focused review paper Stephen and I suggest (in which I agree with Jean that we might have to split it by modality or at the least by types of modalities: anatomical, quantitative, functional etc).

 

That workshop will also help us draft a much more concrete phantom- or QA-focused survey asking specific questions (within an existing framework developed out of the panel discussion).

 

Does that sound like a good idea? If so,

  • Let me what times would be good (to avoid any other major related conferences) – how about late 3rd week of July or early August?
  • in addition to those who chimed in already (with good representation from Canada), I’d like to try get some more inputs from USA and Europe and anywhere else who have done research on this already. So I would appreciate your suggestions for speakers or projects (like ABCD)

 

Thanks,

Pradeep

Yaroslav Halchenko

unread,
Jul 6, 2021, 2:10:58 PM7/6/21
to niQC
On Fri, 02 Jul 2021, Stephen Strother wrote:

>    Another study that I am aware of that is directly relevant to this
>    discussion and is based on a large multi-site set of phantom data is
>    In particular see Fig. 2.

Fig 2 is very nice!  I would have even just collapsed a+c into the scatter plot of e with confidence intervals depicted within the scatter plot, since those "scanning sites" have no order to them anyways if I get it right.

Yet to read it to get a better sense but it feels it would be a great support toward sharing phantom data for a site to support claims that "Brain data" tSNR is quite well explained by phantom tSNR.  I wish we got to analysis of BOLD in our study ;-)

it would have also been cool to see how much of tSNR on this data could be explained with those scanning "parameters" (positioning in the scanner etc) we identified in our paper. 

Oscar Esteban

unread,
Jul 7, 2021, 9:47:24 AM7/7/21
to Yaroslav Halchenko, niQC
Hey Yarik,

I very much support this idea. We have to kick off the acquisition of one "human phantom" on 4 siemens scanners here, with and without personalized headcase. I'll be happy to talk more about this if you'd like.

Cheers,
Oscar

(My postdoc appointment with Stanford has concluded and this account will eventually be closed, please update your contacts - I can be reached at oscar....@unil.ch or p...@oscaresteban.es, thanks!)

___________________________
Oscar Esteban, Ph.D.
Research and Teaching Ambizione FNS Fellow
Dept. of Radiology, CHUV, University of Lausanne

+1 (650) 733 33 82


--
You received this message because you are subscribed to the Google Groups "niQC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to niqc+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/niqc/fc3f3af2-da09-4b7d-b114-e45a3a1203dcn%40googlegroups.com.

Yaroslav Halchenko

unread,
Jul 7, 2021, 11:55:54 AM7/7/21
to niQC
Coolio.  FWIW: don't forget to accompany that poor "human phantom" with some "inhumane phantoms"! ;)

Do you have already your own LEGO phantom?   Here is the baby (with remote assistance by Vladimir Fonov) of our brainhack projects a few years back: http://www.onerussian.com/tmp/emmet-IMG_1028.JPG .  Unfortunately it is still not (yet) a part of the regular QC at our scanner :-/


Simon Duchesne

unread,
Jul 7, 2021, 12:22:07 PM7/7/21
to Yaroslav Halchenko, niQC

Hello Y (and everyone)

 

Am making another concerted push to share all of the CCNA LEGO phantom scans onto the CONP.ca platform for general consumption.

 

Awaiting word from the study leadership to share this inhuman data.

 

Hope it will help

Simon

 

 

From: <ni...@googlegroups.com> on behalf of Yaroslav Halchenko <yarik...@gmail.com>
Date: Wednesday, July 7, 2021 at 11:56
To: niQC <ni...@googlegroups.com>
Subject: Re: [niQC] Re: idea: MRI Clearing House

 

[Externe UL*]

Alexandre Franco

unread,
Jul 8, 2021, 11:50:29 AM7/8/21
to ni...@googlegroups.com
Hi Folks, 

A while back, we presented at EMBC (2014) a poster of a dynamic phantom that we developed. This phantom was developed to try to simulate "BOLD" activation. 
At the time we ended up getting sidetracked and ended up never pursuing this project further, but will be happy to discuss if anyone wants to continue with this idea. 
Here is the link to the poster: 

Best, 
Alex



To view this discussion on the web visit https://groups.google.com/d/msgid/niqc/CAEQhC%2BMY7Fef%2BW%3DRfTtesCrQ-dw0iQGN%2Bto03LOVuSXG9-eG9A%40mail.gmail.com.

Raamana, Pradeep Reddy

unread,
Jul 9, 2021, 10:45:45 AM7/9/21
to Simon Duchesne, Jean Chen, Stephen Strother, ni...@googlegroups.com

Hi Everyone,

 

A gentle reminder to indicate your interest in a short workshop (2-5 hours: few talks on existing work and open discussion) focused on Phantom QA (and discussing various aspects as noted below)?

 

If that’s too much time, perhaps we can start with an open discussion (2 hours) – let me know your availability:

https://doodle.com/poll/dszhhs8dkp6utg49?utm_source=poll&utm_medium=link

 

Thanks,

Pradeep

Raamana, Pradeep Reddy

unread,
Jul 9, 2021, 11:04:50 AM7/9/21
to niQC

Just to clarify, my motivation for this workshop: I believe video recording of these talks and/or panel discussion becomes a valuable educational resource, and they are much easier to participate in (for us) as well as to follow (for others) relative to long threads of emails. Feel free to pushback if you believe emails are doing a fine enough job (asynchronous forums have their advantages).

 

PS: I fixed the doodle poll, it now correctly shows 2 hour chunks

Raamana, Pradeep Reddy

unread,
Jul 9, 2021, 2:11:38 PM7/9/21
to Bandettini, Peter (NIH/NIMH) [E], niQC

No problem at all – I understand this is the summer/travel season (esp. after quarantine life in 2020).

 

I just expanded the poll to include the following week (Jul 26-30) as well to try maximize the attendance.  I just wanted to get this going with those who can join for now, and have more discussions as needed, as it’s unlikely we would solve it all with this one meeting.

 

Here is the link again: https://doodle.com/poll/dszhhs8dkp6utg49?utm_source=poll&utm_medium=link

 

Thanks,

Pradeep

 

From: Bandettini, Peter (NIH/NIMH) [E] <band...@mail.nih.gov>
Date: Friday, July 9, 2021 at 1:59 PM
To: Raamana, Pradeep Reddy <RAAM...@pitt.edu>, niQC <ni...@googlegroups.com>
Subject: Re: [niQC] Re: Phantom QA Meeting

Hi Pradeep,

 

Looking at the poll, I’ll be completely off the grid (hiking/camping in northern New Mexico mountains) during all the times, so can’t make this one.

Sorry about that!

 

Peter

Yaroslav Halchenko

unread,
Jul 9, 2021, 3:03:34 PM7/9/21
to niQC
ALEX, "We should hang out"!!! (reference: Idiocracy)

I was thinking to do 100% the same !  Well, ultimately I hoped to meld it with my "3d printed realistic brain" "project", like this tiny prototype quarter of a slab:
https://twitter.com/yarikoptic/status/1338645010418446336?s=20 (sorry -- can't find locally, but can dig out the dicoms or nifti if interested)
so printing a realistic head(s) with some sensible location for "induced BOLD signal", and making an open Frankenstein dataset/study ;-)

but for the main purpose/question -- that is not needed. If you still have the bucket/liquid etc, my primary question of interest would be -- multiband sequences. It would be fun to see "the functional connectivity of the bucket".

The only thing, I would have reduced SNR in the pumped liquid/voxels to be more inline with a typical BOLD effect -- in your setup it was too different IMHO from the base solution and thus "too clean".

Alexandre Franco

unread,
Jul 9, 2021, 8:48:15 PM7/9/21
to Yaroslav Halchenko, niQC
Hi Yarik, 
Unfortunately, this "phantom" stayed in Brazil and I didn't bring it in my luggage when I moved to the States a few years back. 
I just emailed the folks down there if they still have it (which I doubt). We didn't use it much, this was basically a master's project for one of my students and after he finished we ended up never continuing the project. But will be happy to discuss no how we can revive and improve this dynamic phantom. 

Best, 
Alex




You received this message because you are subscribed to a topic in the Google Groups "niQC" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/niqc/xy51f6t-T6M/unsubscribe.
To unsubscribe from this group and all its topics, send an email to niqc+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/niqc/a772e1d5-c535-4049-8dc5-357f5a54f1ben%40googlegroups.com.

Raamana, Pradeep Reddy

unread,
Jul 16, 2021, 10:24:19 AM7/16/21
to niQC

Gentle reminder everyone to share your interest and availability for an open discussion on phantom QA/QC – by the end of day today. Thanks.

Yaroslav Halchenko

unread,
Jul 16, 2021, 11:35:13 AM7/16/21
to niQC
FWIW, I would be open for any type of collaboration, including internationally ;-)  Unfortunately I have not yet managed to interest anyone locally enough to actually have it done, and can't find time to push on this project/idea alone :-/  So - yeah -- let's stay in touch, might be a nice/cool 2nd-life for the master thesis project.

Yaroslav Halchenko

unread,
Jul 16, 2021, 11:36:11 AM7/16/21
to niQC

Let me know if leadership needs more whining from across the border ;-)

Aina Puce

unread,
Jul 16, 2021, 11:40:24 AM7/16/21
to Yaroslav Halchenko, niQC
Maybe we could develop an co-ordinated initiative like this thru the OHBM Best Practices Committee? It is headed by Peter Bandettini & Jack Van Horn. It was tasked to generate white papers for best practice, but goodness, phantoms are a crucial part of this process too.
This could be really cool - then we could get together ppl who are really interested in this???

To view this discussion on the web visit https://groups.google.com/d/msgid/niqc/8433031c-57b9-45f0-8b50-5cd94d8fd206n%40googlegroups.com.


--
Aina Puce, PhD . 

A button with "Hear my name" text for name playback in email signature 

Eleanor Cox Riggs Professor
Psychological & Brain Sciences
Programs in Neuroscience, Cognitive Science
Affiliate, Indiana University Network Science Institute
Indiana University, Bloomington IN 47405 USA
--
Chair-Past, Organization for Human Brain Mapping

Raamana, Pradeep Reddy

unread,
Jul 16, 2021, 11:53:23 AM7/16/21
to Aina Puce, Yaroslav Halchenko, niQC
Yes Aina - Prof. Jack van Horn and I had a long and productive conversation last Monday and we’re caught up with each other, and we support each other 100%. I was planning to email everyone with some ideas and suggestions I had via this discussion.

One thing I made very clear to Dr. Van Horn, and he agrees with it 100%, is that everyone supports the efforts to develop the QC BPCs/protocols but very few to none are able have the required time and resources to commit to them. 
Sent: Friday, July 16, 2021 11:40:10 AM
To: Yaroslav Halchenko <yarik...@gmail.com>

Yaroslav Halchenko

unread,
Jul 16, 2021, 12:04:24 PM7/16/21
to niQC
+100 on contributing to "OHBM Best Practices"!  Will be waiting for Pradeep's email ;)  It better lists following (from the "expected to be done!" to "requires extra desire/commitment/resources")

- Recommend running regular (at least weakly) MRI QA phantom scans with typical to the center protocols
- Establish (local to the center) archival of the collected MRI QA data
- Make raw MRI QA data available (locally) to interested researchers
- Automate MRI QA and share results publicly
- Share MRI QA "raw" data (ideally publicly)

Aina Puce

unread,
Jul 16, 2021, 12:33:18 PM7/16/21
to Raamana, Pradeep Reddy, Yaroslav Halchenko, niQC
Pradeep - this is cool. So you & I are exactly on the same wavelength here!!! Aina

Henk-Jan Mutsaerts

unread,
Jul 17, 2021, 5:03:42 PM7/17/21
to Raamana, Pradeep Reddy, Simon Duchesne, Jean Chen, Stephen Strother, niQC, Aaron Oliver-Taylor
Very interesting! From the perfusion MRI side, we already have something like this in place with the ISMRM perfusion study group OSIPI (open Science initiative for perfusion imaging), task force 3.2.

Best, Henk

To view this discussion on the web visit https://groups.google.com/d/msgid/niqc/DM6PR04MB6746A876FB080C9C02E00119B8189%40DM6PR04MB6746.namprd04.prod.outlook.com.

Aaron Oliver-Taylor

unread,
Jul 19, 2021, 6:21:39 AM7/19/21
to niQC
Hello All,

I'm from Gold Standard Phantoms (https://goldstandardphantoms.com/), and I came across this group via Henk, and I'm also part of TF 3.2 in OSIPI. Very interesting to have you all discussing this, and also as a phantom manufacturer and we would be very happy to support this.

First of all, regarding a fBIRN phantom, we do indeed make something which is essentially an improved version of the original agar fBIRN phantom - we use a polyacrylamide based gel which has far fewer bubbles, is not a natural product (agar comes from seaweed) so they are more reproducible, and it is made at room temperature (agar must be heated to >85°C, and for the 2.5L or so that goes into one of these spheres it required quite a lot of energy) so more environmentally friendly!  The actual plastic spheres are the exact same ones that have always been used for fBIRN phantoms.  More info here: https://goldstandardphantoms.com/funstar 

Anyway, enough advertising! 

Regarding a centralised place for phantom data to be processed and shared, well we already sort of have something that does this: https://goldstandardphantoms.com/gsp-cloud
Here anyone can create an account to run a fBIRN analysis, along with longitudinal tracking of the results. We can set up groups that multiple users can be part of so for example everyone in that group can see the results from a set of scanners. We also have an analysis on there for ASL data acquired with our perfusion phantom.

A LOT of work went into the initial processing of the DICOMs to correctly identify what the data actually are so that the pipelines can run in an automated manner (in particular for ASL data because this really isn't standardised and a some of the required parameters are missing from the DICOMs). This was by far the largest component over actually implementing the processing pipelines (especially given that something like the fBIRN analysis is very standardised and published).

We offer GSP Cloud as a free tool to support the MR research community, and so I would be very happy to discuss how we could make improvements or even collaborate closely with people from this initiative. 

Best,
Aaron

Aaron Oliver-Taylor

unread,
Jul 19, 2021, 6:26:39 AM7/19/21
to niQC
Slight correction, I'm actually part of TF3.1 in OSIPI: https://osipi.org/task-force-3-1/

The aims of this taskforce is to provide a central point for lists of:
All for perfusion imaging using ASL, DSC or DCE. 

TF3.2 is for an inventory of clinical and preclinical perfusion MRI data: https://osipi.org/task-force-3-2/

Raamana, Pradeep Reddy

unread,
Jul 21, 2021, 1:38:15 PM7/21/21
to Aaron Oliver-Taylor, niQC

Thanks a lot Aaron for sharing GSP and their services in detail. It sounds very useful to our community. We have come across phantomlab.com recently (via Stephen). We can perhaps get you and few other related companies to give a talk on their solutions and services? Hope you can make it for the next week’s meeting.

 

Everyone, I have started to list various commercial solutions here on the niQC website – please feel free to add others you are aware of:

https://incf.github.io/niQC/resources

 

 

 

Registered in England and Wales.

Company number: 9342804

VAT Reg: GB211807049

 

This email and any attachments to it may be confidential and are intended solely for the use of the individual to whom it is addressed. Any views or opinions expressed are solely those of the author and do not necessarily represent those of Gold Standard Phantoms.

If you are not the intended recipient of this email, you must neither take any action based upon its contents, nor copy or show it to anyone.

Please contact the sender if you believe you have received this email in error.

--
You received this message because you are subscribed to the Google Groups "niQC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to niqc+uns...@googlegroups.com.

Aaron Oliver-Taylor

unread,
Jul 22, 2021, 2:56:03 PM7/22/21
to niQC
Unfortunately I'm on holiday next week so can't make the meeting. I'll certainly check out the recording though. If you want to do a follow up event I'd be very happy to talk about what we do there. It would also be a really great opportunity to talk about where there are issues that you all have regarding QA/QC in neuroimaging and find ways we can collaborate. I'm a big believer that a phantom is just one part of the story, for to really be impactful then it needs to be paired with a standardised QA protocol and standardised analysis (the fBIRN test is a great example of this, and also the ACR phantom tests for site accreditation), and while we can design and make the phantoms, we really need collaborate to get the protocol and analyses right (i.e. useful, diagnostic and relevant to the patient/subject data that are being acquired). 

Yaroslav Halchenko

unread,
Dec 1, 2022, 2:14:48 PM12/1/22
to niQC
oh well -- lots of time has gone passed that moment, some folks are even interested to start collecting/sharing (even if only internally within project) phantom QC data, but we never got anywhere with the centralized MRI clearing house idea/pilot setup.
So decided to possibly give it a shot to get some pilot instance running through trying out https://www.orthanc-server.com/  PACs server with https://chrisproject.org/ in tandem with it to provide some processing capabilities.  https://github.com/FNNDSC shows various components which already exist for ChRIS .
So I hope that we could relatively easy establish some setup which would have needed basic components for the clearing house and more (e.g. people doing their own analytics on that data).
I've scheduled to chat with Rudolph Pienaar (AFAIK the tech lead of ChRIS) next Wed 2-3 pm. If interested to join -- let me know and I will add your gmail email address to the guest list for the event ;)

Reply all
Reply to author
Forward
0 new messages