Additional QC Site, UMass...

127 views
Skip to first unread message

Kennedy, David

unread,
Aug 24, 2021, 11:07:56ā€ÆAM8/24/21
to ni...@googlegroups.com, Zheng, Shaokuan, Yaroslav Halchenko
Hi, Shaokuan Zheng at UMass would be willing to consider participation in the QC phantom collection.

I'm looping him in, and not sure the best way to get him up to speed with the plan. Do we have any 'on boarding' for participating sites?

DK

----
David N. Kennedy, Ph.D.

Are we ā€˜making a livingā€™ or ā€˜making a differenceā€™?
"Nothing about [you] without [you]"

Confidentiality Notice:
This e-mail message, including any attachments, is for the sole use of the intendedĀ recipient(s) and may contain confidential, proprietary and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender immediately and destroy or permanently delete all copies of the original message.

Dr Cyril, Pernet

unread,
Aug 24, 2021, 11:41:03ā€ÆAM8/24/21
to Kennedy, David, ni...@googlegroups.com, Zheng, Shaokuan, Yaroslav Halchenko

Hi David,


I think there are two aspects to consider

- phantom only, easy to collect and share

- human/animal data collected by different studies in parallel


happy to adapt what we plan to do as to have multi-site data ..

cyril



--
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/BLAPR10MB5089BBD582CE49D4A0EE144393C59%40BLAPR10MB5089.namprd10.prod.outlook.com.
-- 
Dr Cyril Pernet, PhD, OHBM fellow, SSI fellow
Neurobiology Research Unit, 
Building 8057, Blegdamsvej 9
Copenhagen University Hospital, Rigshospitalet
DK-2100 Copenhagen, Denmark

wamc...@gmail.com
https://cpernet.github.io/
https://orcid.org/0000-0003-4010-4632

Raamana, Pradeep Reddy

unread,
Aug 24, 2021, 3:32:05ā€ÆPM8/24/21
to Dr Cyril, Pernet, Kennedy, David, ni...@googlegroups.com, Zheng, Shaokuan, Yaroslav Halchenko

Iā€™m glad to say weā€™ll be joining this as well. We discussed parts of this during todayā€™s talks, including whether we should consider using one of the newest/latest phantoms presented today, and which metrics/pipelines to apply. Perhaps we can start a google doc and/or have a short meeting to finalize the details before we get started?

Yaroslav Halchenko

unread,
Sep 3, 2021, 12:06:58ā€ÆPM9/3/21
to Bandettini, Peter (NIH/NIMH) [E], ni...@googlegroups.com
Hi Peter,

Do you know if there is some sample/collection of data which was already
collected on such a phantom which could be re-shared/used?

Without looking in detail on phantom operation -- wouldn't it also be
suitable to also assess multiband signal "leakage"?


On Fri, 03 Sep 2021, Bandettini, Peter (NIH/NIMH) [E] wrote:

> I agree with all that was said here. Just to add, there has been work by
> Lili Mujica-Parodi on a dynamic phantom,
> https://alascience.com/products/braindancer-dynamic-phantom/ which, while
> interesting, I think has limited utility as one doesnā€™t need a dynamic
> phantom to estimate time course stability ā€“ but perhaps worth discussing.



> One thing that would be an interesting challenge would be to make a
> phantom that tests local shimming capability ā€“ that has susceptibility
> gradients on the same spatial scale and magnitude as a human head. If you
> really wanted to get fancy, you could try to create a susceptibility
> phantom of a human head so that shims (including construction of a
> wearable passive shim set) might be developed and optimized to counter
> signal dropout where we always see it.

--
Yaroslav O. Halchenko
Center for Open Neuroscience http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
WWW: http://www.linkedin.com/in/yarik

Yaroslav Halchenko

unread,
Sep 3, 2021, 3:22:46ā€ÆPM9/3/21
to LR Mujica-Parodi, Helmut Strey, ni...@googlegroups.com, Bandettini, Peter (NIH/NIMH) [E]
Dear Lily and Helmut,

We have made a full "circle" and got back again to your interesting
phantom. (I am following up here on a "public" thread within niQC
group I have mentioned to your before).

Please let us know if you need any assistance with depositing the data
for that study (https://doi.org/10.1016/j.neuroimage.2020.117584) to
OpenNeuro.

But even more interesting (to me) is either you had any additional
multiband acquisitions. If I glanced through your paper "correctly" --
that data was not multiband; and that-is-partially-why I was
reaching to see the data's sidecar files to possibly find more details
on acquisitions.

Cheers,

On Fri, 03 Sep 2021, Bandettini, Peter (NIH/NIMH) [E] wrote:

> Thanks Yaroslav, yes this might be interesting to asses how multi-band
> varies over time with something that varies dynamically ā€“ assuming that
> the change in the phantom somehow might have an influence on the recon for
> each image.



> If you contact Lili directly, she could direct you do the person who has
> been accumulating data. Hereā€™s a relevant paper
> https://pubmed.ncbi.nlm.nih.gov/33285328/ from their group. Rajat Kumar
> was leading this study and collecting data.



> Peter





> From: Yaroslav Halchenko <yarik...@gmail.com>
> Date: Friday, September 3, 2021 at 12:08 PM
> To: Bandettini, Peter (NIH/NIMH) [E] <band...@mail.nih.gov>
> Cc: ni...@googlegroups.com <ni...@googlegroups.com>
> Subject: Re: [niQC] Additional QC Site, UMass...

Raamana, Pradeep Reddy

unread,
Sep 3, 2021, 3:26:39ā€ÆPM9/3/21
to Yaroslav Halchenko, LR Mujica-Parodi, Helmut Strey, ni...@googlegroups.com, Bandettini, Peter (NIH/NIMH) [E]

Thanks Peter and Yarik ā€“ this will make for an interesting talk and discussion. Helmut and Lily were willing to give a talk, I put it on hold to try group with very related talks from Dr. Joshua Kupermanā€™s of ABCD as well as potentially another from Dr. Gary Glover. Perhaps we should wait no longer and do it in the next few weeks when it works for all those interested?

Ā 

From: ni...@googlegroups.com <ni...@googlegroups.com> on behalf of Yaroslav Halchenko <yarik...@gmail.com>
Date: Friday, September 3, 2021 at 3:22 PM
To: LR Mujica-Parodi <liliann...@stonybrook.edu>, Helmut Strey <helmut...@stonybrook.edu>
Cc: ni...@googlegroups.com <ni...@googlegroups.com>, Bandettini, Peter (NIH/NIMH) [E] <band...@mail.nih.gov>
Subject: Re: [niQC] Additional QC Site, UMass...

Dear Lily and Helmut,

We have made a full "circle" and got back again to your interesting
phantom. (I am following up here on a "public" thread within niQC
group I have mentioned to your before).

Please let us know if you need any assistance with depositing the data


OpenNeuro.Ā 

But even more interesting (to me) is either you had any additional
multiband acquisitions. If I glanced through your paper "correctly" --
that data was not multiband; and that-is-partially-why I was
reaching to see the data's sidecar files to possibly find more details
on acquisitions.

Cheers,

On Fri, 03 Sep 2021, Bandettini, Peter (NIH/NIMH) [E] wrote:

>Ā Ā Ā  Thanks Yaroslav, yes this might be interesting to asses how multi-band
>Ā Ā Ā  varies over time with something that varies dynamically ā€“ assuming that
>Ā Ā Ā  the change in the phantom somehow might have an influence on the recon for
>Ā Ā Ā  each image.



>Ā Ā Ā  If you contact Lili directly, she could direct you do the person who has
>Ā Ā Ā  been accumulating data. Hereā€™s a relevant paper


>Ā Ā Ā  was leading this study and collecting data.



>Ā Ā Ā  Peter





>Ā Ā Ā  From: Yaroslav Halchenko <yarik...@gmail.com>
>Ā Ā Ā  Date: Friday, September 3, 2021 at 12:08 PM
>Ā Ā Ā  To: Bandettini, Peter (NIH/NIMH) [E] <band...@mail.nih.gov>
>Ā Ā Ā  Cc: ni...@googlegroups.com <ni...@googlegroups.com>
>Ā Ā Ā  Subject: Re: [niQC] Additional QC Site, UMass...

>Ā Ā Ā  Hi Peter,

>Ā Ā Ā  Do you know if there is some sample/collection of data which was already
>Ā Ā Ā  collected on such a phantom which could be re-shared/used?

>Ā Ā Ā  Without looking in detail on phantom operation -- wouldn't it also be
>Ā Ā Ā  suitable to also assess multiband signal "leakage"?

>Ā Ā Ā  On Fri, 03 Sep 2021, Bandettini, Peter (NIH/NIMH) [E] wrote:

>Ā Ā Ā  >Ā Ā Ā  I agree with all that was said here. Just to add, there has been work
>Ā Ā Ā  by
>Ā Ā Ā  >Ā Ā Ā  Lili Mujica-Parodi on a dynamic phantom,


>Ā Ā Ā  while
>Ā Ā Ā  >Ā Ā Ā  interesting, I think has limited utility as one doesnā€™t need a
>Ā Ā Ā  dynamic
>Ā Ā Ā  >Ā Ā Ā  phantom to estimate time course stability ā€“ but perhaps worth
>Ā Ā Ā  discussing.

>Ā Ā Ā  >Ā Ā Ā  One thing that would be an interesting challenge would be to make a
>Ā Ā Ā  >Ā Ā Ā  phantom that tests local shimming capability ā€“ that has
>Ā Ā Ā  susceptibility
>Ā Ā Ā  >Ā Ā Ā  gradients on the same spatial scale and magnitude as a human head. If
>Ā Ā Ā  you
>Ā Ā Ā  >Ā Ā Ā  really wanted to get fancy, you could try to create a susceptibility
>Ā Ā Ā  >Ā Ā Ā  phantom of a human head so that shims (including construction of a
>Ā Ā Ā  >Ā Ā Ā  wearable passive shim set) might be developed and optimized to
>Ā Ā Ā  counter
>Ā Ā Ā  >Ā Ā Ā  signal dropout where we always see it.
--
Yaroslav O. Halchenko


Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755


--
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.

Raamana, Pradeep Reddy

unread,
Sep 10, 2021, 9:51:15ā€ÆAM9/10/21
to Dr Cyril, Pernet, Kennedy, David, ni...@googlegroups.com, Zheng, Shaokuan, Yaroslav Halchenko

Hi Cyril, Yarik, David, Patrick and others interested,

Ā 

Should we try finalize the protocol and other details for this multi-site phantom data collection? Google doc and/or a short meeting to start with?

Ā 

Thanks,

Pradeep

Dr Cyril Pernet

unread,
Sep 10, 2021, 10:12:16ā€ÆAM9/10/21
to Raamana, Pradeep Reddy, Kennedy, David, ni...@googlegroups.com, Zheng, Shaokuan, Yaroslav Halchenko

sure - let's start a google doc . once we have our thoughts/protocol prpoposal down, we can zoom

-- 
Dr Cyril Pernet, PhD, OHBM fellow, SSI fellow
Neurobiology Research Unit
Copenhagen University Hospital, Rigshospitalet
Building 8057, Blegdamsvej 9
DK-2100 Copenhagen, Denmark

Yaroslav Halchenko

unread,
Sep 10, 2021, 10:22:29ā€ÆAM9/10/21
to ni...@googlegroups.com
FWIW: thumbs up

On Fri, 10 Sep 2021, Dr Cyril Pernet wrote:

> sure - let's start a google doc . once we have our thoughts/protocol
> prpoposal down, we can zoom

--
Yaroslav O. Halchenko
Center for Open Neuroscience http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
WWW: http://www.linkedin.com/in/yarik

Dr Cyril, Pernet

unread,
Sep 10, 2021, 3:12:40ā€ÆPM9/10/21
to Yaroslav Halchenko, ni...@googlegroups.com
left my thoughts and ideas on the docĀ  - I will get the exact MRI
parameters next week

cheers


On 10/09/2021 16:22, Yaroslav Halchenko wrote:
> FWIW: thumbs up
>
> On Fri, 10 Sep 2021, Dr Cyril Pernet wrote:
>
>> sure - let's start a google doc . once we have our thoughts/protocol
>> prpoposal down, we can zoom

--

Open Minds Lab

unread,
Sep 10, 2021, 4:02:46ā€ÆPM9/10/21
to Dr Cyril, Pernet, Yaroslav Halchenko, ni...@googlegroups.com
Here is the link to a rough draft -- please feel free to share your comments and suggestions: https://docs.google.com/document/d/1IZj-CdIJHbiBDV8Uy9fGa_s-bW4V-5jnavCFo2TgxD0/edit?usp=sharing

Ā 

--
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/497256fa-8269-b067-8412-c74116c9d07b%40gmail.com.
Message has been deleted

Open Minds Lab

unread,
Oct 6, 2021, 3:36:33ā€ÆPM10/6/21
to John Pyles, niQC
Fabulous šŸ˜€.. Welcome, John! You arenā€™t late at all - weā€™re about to schedule a meeting shortly and finalize theĀ details to get started on it.Ā 

On Tue, Oct 5, 2021 at 5:35 PM John Pyles <jo...@uw.edu> wrote:
Hi all,

I'm a little late to the party as I just found out about this group, but I would love to participate in the phantom data collection. We're very interested in facility QA at our new center at Univ. of Washington, and I think this cross site collection is a great idea. Besides the standard Siemens phantoms we also have the GSP FUNSTAR and SPECTRE available.

I'm also actively trying to figure out a good solution for holding spherical phantoms if there are ideas for this (I'll also comment in the Google document). My previous method of using custom milled holders from CaseForge ceased to be an option when they shut down in the pandemic. There is a thread about this over in the Siemens Neuro User Group discussion board (https://www.snug-discuss.org/) I run with Ben Inglis. Any suggestions there or here would be great. My current goal is to try to get some good 3D scans of the standard Siemens headcoils (and other manufacturers too?).Ā 

Looking forward to participating in any future meetings and this effort!Ā 

Best,
John

__________________________________________________

John Pyles, Ph.D.
Associate Director, CHN
Principal Research Scientist
Center for Human Neuroscience
Department of Psychology
University of Washington
email: jo...@uw.edu Ā | Ā phone: 206.552.0107
__________________________________________________

Yaroslav Halchenko

unread,
Oct 6, 2021, 3:45:59ā€ÆPM10/6/21
to ni...@googlegroups.com

On Tue, 05 Oct 2021, John Pyles wrote:
> I'm also actively trying to figure out a good solution for holding
> spherical phantoms if there are ideas for this (I'll also comment in the
> Google document). My previous method of using custom milled holders from
> CaseForge ceased to be an option when they shut down in the pandemic.

may be we should create parametric 3D model to just 3d print it to
support such phantoms in a known/fixed position within the coil?

IMHO we could make it nice and tight within any particular head coil.

Aaron Oliver-Taylor

unread,
Oct 7, 2021, 6:10:49ā€ÆAM10/7/21
to niQC

Hi All,

Every now and again we get some requests about coil-specific holders for the FUNSTAR/SPECTRE phantoms, but it's not really worthwhile for one-offs. However, if there is significant interest from the community then we would be keen to develop and release phantom holders for various coils.Ā  The questions I have are:

1. Is there a quantifiable advantage to using a phantom holder - has this been demonstrated anywhere for a fBIRN test? Or any other types of QC
2. Even if there isn't any benefit in terms of QA results, does the ease of setup justify some of the value alone?
3. How much would people be willing to pay for such a holder - we're a company so of course need to cover our costs if we offer these. 3D printing is of course an option for those with printers, but it will likely be quite large which means potential failed prints.Ā 
4. It would also be good to know which coil(s) you would want a hold for.

Cheers,
Aaron

romain

unread,
Oct 7, 2021, 6:21:13ā€ÆAM10/7/21
to ni...@googlegroups.com

Hello,

I do not have that much experience, but from the few test I did on a phantom it was indeed very important, to reproduce the exact same position of the phantom

I guess this will be dependent on the specific metric you use, but for SNR, there is a strong dependency of Signal level if you are close or not to the coil.

So mean SNR in your phantom will vary a lot for different position of the phantom relative to the coil ...

so if you want to look at SNR fluctuation through time, you have to ensure a reproducible phantom positioning ...


Romain



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.

Christoph Vogelbacher

unread,
Oct 7, 2021, 8:32:46ā€ÆAM10/7/21
to niQC

Hey all,Ā 

I created a phantom holder for a gel phantom and there is definitvly a benifit for it (https://figshare.com/articles/poster/Quality_Assurance_for_functional_Magnetic_Resonance_Imaging/5817486). I use this holder for the QA of our scanner. As well we created a holder for the ACR phantom. The holder enables the possibility to align the phantom always the same way inside the headcoil and a fix protocol to position the measurement volume can be used. Also the time to align the phantom was redused.Ā 

I don't know how much a person is willed to pay for a holder (typically nothing :-) ) but maybe we can provide printerfiles or some blueprints.Ā 

We also have to test if the printed holder makes any unwanted artifacts.Ā 

We do not only need the information which head coil is used we have to consinder which phantom we want to use. so its the combination phantom x headcoil.

Raamana, Pradeep Reddy

unread,
Oct 7, 2021, 8:51:07ā€ÆAM10/7/21
to Christoph Vogelbacher, niQC

Good point Christoph ā€“ perhaps the lack of significant differences between Manual Alignment and Phantom Holder was because you put in 5-10 mins to align it as best you as you could. If one were to not pay much attention or effort to it, weā€™d very likely see significant differences.

Ā 

This makes wonder how often people use phantom holders? Among the subset doing QA, that is šŸ˜Š Ive a feeling itā€™s not the majority.

Ā 

Which leads me to an obvious best practice recommendation for QA:

Use phantom holders, reduce variability, improve QA and save time! #YOLO

Ā 

Looking fwd to the Oct 21 panel to learn what different centres do.

Ā 

From: ni...@googlegroups.com <ni...@googlegroups.com> on behalf of Christoph Vogelbacher <christophv...@gmail.com>
Date: Thursday, October 7, 2021 at 8:34 AM
To: niQC <ni...@googlegroups.com>
Subject: Re: [niQC] Additional QC Site, UMass...


Hey all,Ā 

John Pyles

unread,
Oct 7, 2021, 6:01:11ā€ÆPM10/7/21
to niQC
Hi all,

Thanks for the welcome! I do not have specific data I've collected myself that shows the importance of consistent positioning, so thanks Romain and Christoph for sharing your experiences. It definitely makes theoretical sense to me that you'd want the phantom to be a consistent distance away from the various receive coils across the different measurements you are comparing. I'd be happy to collect additional data on this if it is of interest?

Practically speaking, a holder is also good for making setup easy and consistent between different people running the QA. Holders are already provided for the factory phantoms, so I think it would be great to have holders for spherical phantoms such as FUNSTAR and fBIRN.

To answer Aaron's question: I think the price I (and possibly others) would be willing to pay is something around what the cost would be to 3D print the holder ourselves, plus a reasonable markup. An optimal solution to me would be to have publicly available 3D print files, but then also the ability to just buy already printed holders from a company like GSP if the means or time are not available to print your own.Ā 

I am not a 3D printing expert at all, but the first step seems to be to get high resolution 3D scans of the head coils we'd want holders for? For Siemens this would be the 20ch, 32ch, and 64ch head and head/neck coils. I don't know what the GE and Philips coils are in common use right now, but my guess is that it is probably 2-3 coils per manufacturer. So that puts us at around 9 different coils total. While some work, that seems potentially reasonable to me to have 9 holder designs. The fBIRN, FUNSTAR, and SPECTRE I believe are all the same size, so that would mean one holder could cover a number of phantoms.Ā 

I'm looking into getting 3D scans of Siemens and potentially some Philips coils here. But if those scans already exist that would be great.

--John

Yaroslav Halchenko

unread,
Oct 7, 2021, 8:34:45ā€ÆPM10/7/21
to ni...@googlegroups.com

On Thu, 07 Oct 2021, John Pyles wrote:
> I am not a 3D printing expert at all, but the first step seems to be to
> get high resolution 3D scans of the head coils we'd want holders for? For
> Siemens this would be the 20ch, 32ch, and 64ch head and head/neck coils. I

we should have a good collection of caseforge cases for our siemens 64ch
coil. I guess they are a good fit. Will try to measure and/or scan
them next week to get some prelim model.

The tricky part for 3d printable model would be to create it with nice
joints or modular elements (we do not really need "full enclosure" --
just a nice tightly fit holder) since printer wouldn't fit the
entire holder.

> don't know what the GE and Philips coils are in common use right now, but
> my guess is that it is probably 2-3 coils per manufacturer. So that puts
> us at around 9 different coils total. While some work, that seems
> potentially reasonable to me to have 9 holder designs. The fBIRN, FUNSTAR,

100%

and most likely their shape is quite "parametric" and, again, we don't
need a full "space"/enclosure (e.g we don't care about filling the
top space) -- just a nice cradle to position phantom on and
unambigously slide/position into the head coil for scanning.

Aaron Oliver-Taylor

unread,
Oct 8, 2021, 5:26:58ā€ÆAM10/8/21
to niQC
We have models for the Siemens coils (64/32/20ch - you can get them on request after signing a NDA with Siemens, so unfortunately I cannot share these with anyone),Ā  Philips 32ch coil (based on our measurements), and the GE 32ch coil (again based on our measurements). And, of course CAD design and manufacturing things that go in the MRI scanner is our bread and butter.

I'm not the biggest fan of 3D printing to be honest (and I write this sitting no more than 3ft away from a Creality Ender 3...) - while the desktop 3D printers have improved dramatically over the years, printing large objects is still a pain and subject to failure so in reality it might not be as cheap and easy as it sounds. And yes it would most likely need to be printed in several parts and then joined together. I agree it's nice to have a viable 'open' solution that those who are that way inclined can make use of.

There are other manufacturing technologies that are far cheaper per part (injection moulding, foam casting etc) that would keep the cost down of something that is sold. But of course they require a reasonably significant investment to begin with - making moulds etc.

Are the holders that come with the MRI vendor phantoms coil specific, or are they merely just a cradle for the phantom? I'm wondering how necessary it is to stratify by coil - even if it doesn't have perfectly conforming curves a single design for all coils might do the job.Ā 

John Pyles

unread,
Oct 9, 2021, 3:20:59ā€ÆPM10/9/21
to niQC
Hi Aaron,

Thanks for the additional info and thoughts. That's great that you have the Siemens coil models! I didn't know they would provide them (even with an NDA). The Siemens holder for their "bullet" phantom is a foam rubber holder that works in the 20, 32, and 64ch coils. I'd say the positioning is okay, not great. Probably within a 5-10mm. It is a cradle design though that works well for a cylindrical phantom, but would not really work for a sphere. I'll take a photo next time I'm at our magnet.

An idea for a potentially cheap and easy(ish) to make holder -- What about a holder based on a two-part interlocking flower pot stand? (see attached photos) The top holds the spherical phantom, and the bottom of the two pieces are cut to conform exactly to the specific coil the holder is designed for? The two parts could be easily cut out of plywood or acrylic on a CNC router or cutter. Then you just have one point to glue them together for assembly. The bottom edges that touch the coil and the top that touch the phantom could be covered in thin neoprene strippingĀ for some added protection. No 3D printing and only pretty basic CNC work. What do you think?

--John

PotStand.jpgStand-Parts-1.JPGStand-Parts-2.JPG

Raamana, Pradeep Reddy

unread,
Oct 9, 2021, 3:50:13ā€ÆPM10/9/21
to John Pyles, niQC

I was also thinking about ā€œinflatableā€ holders (that could diversity of shapes etc) but there might be other issues (physics, sturdiness, stability etc) with such an idea.

Ā 

From: ni...@googlegroups.com <ni...@googlegroups.com> on behalf of John Pyles <jo...@uw.edu>
Date: Saturday, October 9, 2021 at 3:21 PM
To: niQC <ni...@googlegroups.com>
Subject: Re: [niQC] Additional QC Site, UMass...

Hi Aaron,

Ā 

Thanks for the additional info and thoughts. That's great that you have the Siemens coil models! I didn't know they would provide them (even with an NDA). The Siemens holder for their "bullet" phantom is a foam rubber holder that works in the 20, 32, and 64ch coils. I'd say the positioning is okay, not great. Probably within a 5-10mm. It is a cradle design though that works well for a cylindrical phantom, but would not really work for a sphere. I'll take a photo next time I'm at our magnet.

Ā 

An idea for a potentially cheap and easy(ish) to make holder -- What about a holder based on a two-part interlocking flower pot stand? (see attached photos) The top holds the spherical phantom, and the bottom of the two pieces are cut to conform exactly to the specific coil the holder is designed for? The two parts could be easily cut out of plywood or acrylic on a CNC router or cutter. Then you just have one point to glue them together for assembly. The bottom edges that touch the coil and the top that touch the phantom could be covered in thin neoprene strippingĀ for some added protection. No 3D printing and only pretty basic CNC work. What do you think?

Ā 

--John

Ā 

--

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.

John Pyles

unread,
Oct 9, 2021, 4:02:37ā€ÆPM10/9/21
to niQC
Hi Pradeep,

Thanks! An "inflatable" holder is actually similar to what I'm currently using, but a slight improvement (I think you'd definitely setup vibrations with an air based holder). We've been using "holders" made from Instapak chemical foam packing bags. See photos below. You can mold the foam to the coil and phantom and then it hardens. This is actually what the company that currently makes the fBIRN recommends. It works okay, but the precious I don't think is great, and the foam is not durable enough to last long term. It is also impossible to exactly replicate -- there is no way you could create the same foam holder twice with the same thicknesses on the sides and bottom. So that means you cannot have the exact same position between sites which I think is important for cross center data sharing.Ā 

--John

FoamHolder-1.jpgFoamHolder-2.jpg

Raamana, Pradeep Reddy

unread,
Oct 9, 2021, 4:14:51ā€ÆPM10/9/21
to John Pyles, niQC
Yeah, thanks John - I agree with you, and would have been surprised if none of the smart physicists thought of that idea šŸ˜„

can someone put rough $$ numbers to those holders, 3D print and injection moulding types? I feel, given they are the best practice and are needed long term, we should go with the most durable and table, over a period of 10 years? I think weā€™re still using an MRS phantom here from 1997 IIRC šŸ˜€

Sent: Saturday, October 9, 2021 4:02:37 PM

Aaron Oliver-Taylor

unread,
Oct 12, 2021, 7:55:10ā€ÆAM10/12/21
to niQC
OK let's take an example of something that could be made for the Siemens 32 channel coil. I've just knocked up something, it's not ideal at all (and I wouldn't be the one designing something if we were to do this properly!):

32ch.PNGsphere+holder.PNGholder.PNGcross-section.PNG

Here's the 32ch coil with a 18cm sphere (FUNSTAR/SPECTRE/fBIRN phantom) positioned in the middle of the coil (RF centre H-F direction, and at the mid-point of the coil A-P direction). I've created a very simple 'holder' by revolving a shape that confirms to the inside of the coil, and subtracting the sphere to make an indentation for it to sit in.Ā  The holder has a volume of 1560 cm3.Ā 

3D print
Let's assume it's printed at 20% infill with PLA (density ~1.25g/cm3) -> 390g of filament. A 1kg reel is about Ā£15, so Ā£5.85 worth of material. It might take 24 hours to print, let's say a desktop 3D printer is drawing 500W throughout the print (likely to be less) and let's say electricity costs 25p/kWh. So that is Ā£3 of electricity - again likely an overestimation.
Assuming your printer is well set up so the print does not fail, and is large enough to handle this in the first instance, and you only count the material + electricity it will cost about Ā£9.

Let's be conservative and say that it costs 4x that because it fails a few times - Ā£36. As for durability I'm not sure - these phantoms weigh close to 3kg each, and 20% infill is not a lot. All it takes is a little bit of delamination and the holder will start to fall apart, and in my experience this does happen with PLA. So, let's say that you need to re-print once per year - total cost over 10 years is Ā£360, not counting labour or the fact that the person who knew how to use the 3D printer and get it working nicely has left the lab... (yes I'm being a bit cynical)

If you outsource to a service it will cost more (because they are running a business) and of course there will be shipping costs.

CNC/Laser Cut
The costs here are dependent on the material, and of course access to the machine. A design would need to be made that comprises two planar pieces that then slot together.Ā  Laser cutting would be really quite simple - this could be made from acrylic or another laser-cuttable plastic, or potentially something like EVA or polyurethane foam. Due to the conical beam, and variable kerf the reproducibility of this might not be great between different machines. Of course outsourcing to a professional company who are manufacturing to a drawing would be very different, but of course will cost more and is only worthwhile in volume.

A CNC router/mill brings in the issue of workholding, but will be more dimensionally accurate and have parallel edges. 'Programming' a CNC is more time consuming than it is for laser cutting or 3D printing, so this + general setup means there would be additional costs (if outsourced), which again means this is more worthwhile in volume.

So long as they are designed well, parts made using this technique should be very durable over a long period of time.

Injection Moulded
The design would need to be adjusted significantly to something that can be moulded (thin walls + ribs instead of solid sections), but this is very doable. Then there is an initial outlay of getting moulds fabricated which would be several thousand Ā£. After that, for every run there are setup fees (few hundred Ā£) and then the price per part is cheap (few Ā£). The parts will be very strong and durable, have excellent dimensional reproducibility, and there will be the opportunity for the widest range of materials.

Here though, it is only worthwhile if a large number of people want to purchase these designs. Let's say (these are reasonable estimates from things we have injection moulded in the past):

Engineering work: Ā£1000
Mould construction: Ā£4000
Setup Fee: Ā£400
Price per part: Ā£5

So let's say we have an initial run of 100 parts - that's Ā£5900 in total - effectively Ā£59 per part. For a run of 200 it's Ā£32 per part - but will we really sell 200 that quickly or will they just sit in a box for a few years?!

There are other techniques potentially such as foam casting which would probably sit somewhere in the middle in terms of cost.

I would also want to try to have something that is modular, with documented hole patterns etc so that users could 3D print their own attachments to support different phantoms (or other things they want to fix inside the scanner).

All-in-all though, the actual parts cost only represent a fraction of the total commercial cost of offering a product like this, so if I were to pluck a number out of the air as a general feeling for what the end cost for having commercially available phantom holders that are reproducible and anyone can easily buy them then we would probably want to be charging about 10% of the cost of the phantom - so around Ā£200. Does that sound fair?Ā 

John Pyles

unread,
Oct 13, 2021, 2:09:27ā€ÆAM10/13/21
to niQC
Hi Aaron,

This is great! Thanks so much for taking the time for the thorough summary and quick design! A few thoughts...

I actually think the design you came up with quickly is very good! It achieves the goal simply, is a simple shape, and meets two of the main requirements I have for a holder.Ā 
  • First being the phantom is easy to get in and out. Because you've done a partial cradle, this should be no problem at all. The issue with my holders from CaseForge is that they were full half spheres, which made it quite difficult to get the phantom in and out because of resistance and no place to grab the sides of the phantom. This design solves that issue.
  • Second is replicable placement in the coil. There might be room for improvement, but this seems pretty good to me. Placement would involve just two steps: getting the holder flush with the back of the coil, matching up a centerline mark in the holder and coil. Done.

That being said, it seems like there are challenges to production 3D shapes - either some relatively advanced 3D printing is needed, or expensive injection molding setup. This make me lean towards a CNC/Laser Cut design in two pieces that slot together. It seems like this would be the easiest, simplest, and cheapest solution. I think a lot of universities now have maker spaces with CNC routers and/or laser cutters available. So it seems like open source plans could allow these to be made locally fairly easily. Very reasonably priced commercial CNC services also seem easy to find. Taking a quick look there appears to be an online service that could make plywood cuts for maybe $5-10 a part, and there are at least three local shops with CNC, laser, and waterjet cutting offering services here in Seattle.Ā 

Regarding cost -- I believe I paid around $150 each for the CaseForge phantom holders. I wasn't charged any design fees though. So Ā£200 is a bit more expensive, but we're still talking under $1000 for a set of three holders (20, 32, 64ch). That to me seems well worth it given the price Siemens charges for pads and replacement foam rubber holders, and the cost of MRI peripherals in general. Not sure what others think?

What are your thoughts at this point? I'd be happy to pay GSP the quoted price per holder. Or another option might be to pay you for your design services since you have the Siemens CAD files, and then make the design open source. I'd also be happy to contribute to a commercial run of a 2D (or even 3D) design.Ā 

Great discussion! I think we're making progress!Ā 

--John
Reply all
Reply to author
Forward
0 new messages