I would create that spreadsheet in Google Docs, and link from here. We can also create a Resource entry in this community, and link to the google spreadsheet.
To monitor updates, there is a mechanism in Google Docs to send email updates every time someone makes a change to the document (see menu Tools > Notification rules).
The checked tags that are utilized are marked by "x", and not found - by "NA". The "standard" DICOM tags from DWI macro-document are included for all vendors (even though they are not used by GE and SIEMENS). Interestingly none of the vendors utilized the "required" (?)diffusion sequence tag (0018,9117).
So far, there are three different SW versions listed for each vendor. Please, add more, as you come across new version/s for your scanner/s.
Dariya, thanks - this is a great start! I didn't respond earlier, since I accidentally turned off subscription to this thread, and do not know how to re-enable it! :(
I have few suggestions
1) I found the spreadsheet to be a bit difficult to read, so I rearranged stuff a bit - can you take a look at the sheet I added?
2) what do you think about initiating a public repository of de-identified sample datasets to accompany each of the combinations? This would be very helpful for developing a suite of conversion tests down the road. We have some in the Slicer community that I should be able to contribute.
1) data sharing: I suggest we create a shared folder in Google Drive to share those alongside the spreadsheet; TCIA has a long approval process, and interest in phantoms may be limited - this should not interfere with our objectives. I created a folder on Google Drive: http://goo.gl/xb1IjG. Please request write access by sending me a direct message (I will add Dariya and Steve). If space becomes an issue, there are alternatives.
2) thanks for the discussion re cells meaning - I haven't actually tried to make sense of those, just formatting for better comprehension (we should harmonize with other sources listed in https://nciphub.org/groups/qindicom/wiki/DiffusionrelatedDICOMtags:experienceacrosssites); for now, I added another spreadsheet "Legend" to include the descriptions
@Jayashree: I see the following advantages of using GoogleDrive:
1) easy synchronization with the desktop
2) more intuitive interface
3) easy access control beyond nciphub community
4) as I've heard, GoogleDrive can be integrated in some way with nciphub, so in a way the answer could be - "sure, it is already on nciphub" without any extra work
5) consistent with our use of GoogleDrive for the attributes spreadsheet - I guess we could use nciphub databases (if this is the right name) for that purpose as well, but I do not see the added benefit, and the process of editing is rather cumbersome
By the way, are there any existing examples of utilizing nciphub for a similar purpose?
We have already attempted to include these sources (wiki and converters) into the summary that we have. In lines with QIN ADC challenge, this discussion focuses on DWI (*not* DTI), so DTI attributes are omitted on purpose.
>> If you detect missing DWI attributes, please edit/add as needed (by following the link to Google docs), as long as you have a DWI-DICOM data example to support your entry.
>>Current spreadsheet has the list of attributes for which we have DWI-DICOM examples (to be uploaded soon).
>>This library is just the "seed", and we hope and invite more data to be added by all sites that have different MRI systems and SW versions.
We included standard tags for all vendors in the spreadsheet -- they are mostly "NAs"... Philips seems the closest to standard DICOM for "critical" DWI info (although some tags are still missing, as you can see from examples).
Our hope is that QIN BIDS group will help us with vendor-agnostic DWI conversion to *standard* DICOM by mid-summer. For their reference, we are creating this DICOM tag library across vendors. This will also be related back to vendors through QIN and RSNA.
Dariya, to clarify, we discussed providing tools to save ADC maps in a DICOM-compliant format. Producing a tool to harmonize all these vendors into a standard-compliant shape is beyond what we can accomplish in QIICR. We can discuss this later, when we are better prepared with ADC tools.
No matter what, the work of putting together this spreadsheet and developing a collection of sample data will be a very valuable resource.
Andrey Fedorov @ on — Edited @ @ on
Folks, the wiki page has been created: https://nciphub.org/groups/qindicom/wiki/DiffusionrelatedDICOMtags:experienceacrosssites. Let's discuss the next steps in this thread.
Report abuse
Dariya Malyarenko @ on
To monitor consistency of DWI DICOM tags across vendors and SW versions, we could start a library/collection, e.g., as a spread-sheet summary?
Report abuse
Andrey Fedorov @ on
I agree, a single spreadsheet summarizing information from various sources would be helpful.
Report abuse
Dariya Malyarenko @ on
I would start the library (xsl) -- where should we keep it and how to monitor the updates?
Report abuse
Andrey Fedorov @ on
I would create that spreadsheet in Google Docs, and link from here. We can also create a Resource entry in this community, and link to the google spreadsheet.
To monitor updates, there is a mechanism in Google Docs to send email updates every time someone makes a change to the document (see menu Tools > Notification rules).
Report abuse
Dariya Malyarenko @ on
I created the initial library for SW-version versus utilized DWI DICOM tags, as suggested (separate tabs for each vendor):
https://docs.google.com/spreadsheets/d/16RmC4dSAM6-H0MCqh8IumAtPMmOGOmftN7lsZrEecS8/edit?pli=1#gid=1261037047
The checked tags that are utilized are marked by "x", and not found - by "NA". The "standard" DICOM tags from DWI macro-document are included for all vendors (even though they are not used by GE and SIEMENS). Interestingly none of the vendors utilized the "required" (?)diffusion sequence tag (0018,9117).
So far, there are three different SW versions listed for each vendor. Please, add more, as you come across new version/s for your scanner/s.
Report abuse
Andrey Fedorov @ on
Dariya, thanks - this is a great start! I didn't respond earlier, since I accidentally turned off subscription to this thread, and do not know how to re-enable it! :(
I have few suggestions
1) I found the spreadsheet to be a bit difficult to read, so I rearranged stuff a bit - can you take a look at the sheet I added?
2) what do you think about initiating a public repository of de-identified sample datasets to accompany each of the combinations? This would be very helpful for developing a suite of conversion tests down the road. We have some in the Slicer community that I should be able to contribute.
Report abuse
Andrey Fedorov @ on
Also, you abbreviated some names, and I am not exactly sure about the meaning ... please see "Philips DWI reorganized" sheet in google doc.
Report abuse
Andrey Fedorov @ on
I added transposed sheets for the other 2 sheets.
About the data, it is very important that we have the DICOMs. Text files are not possible to process with software, so they are of very limited use.
If the images are phantom images, I do not see any issues making them public. There are no PHI concerns.
Report abuse
Steve Pieper @ on
I like the spreadsheet! Can you clarify the meaning of the entries in the cells though?
Here's what I guess based on what I see:
x : tag is reliably available
x/x/x : three tags are available
NA : not available
x(?iso) : I don't know
x(HS) : I don't know
x(+10^6) : need to add a million to the value?
x(FD?) : I don't know
Thanks in advance for clarifying these.
Report abuse
Andrey Fedorov @ on
1) data sharing: I suggest we create a shared folder in Google Drive to share those alongside the spreadsheet; TCIA has a long approval process, and interest in phantoms may be limited - this should not interfere with our objectives. I created a folder on Google Drive: http://goo.gl/xb1IjG. Please request write access by sending me a direct message (I will add Dariya and Steve). If space becomes an issue, there are alternatives.
2) thanks for the discussion re cells meaning - I haven't actually tried to make sense of those, just formatting for better comprehension (we should harmonize with other sources listed in https://nciphub.org/groups/qindicom/wiki/DiffusionrelatedDICOMtags:experienceacrosssites); for now, I added another spreadsheet "Legend" to include the descriptions
Report abuse
Jayashree Kalpathy-Cramer @ on
Regarding 1: why not use nciphub itself?
Report abuse
Andrey Fedorov @ on
@Jayashree: I see the following advantages of using GoogleDrive:
1) easy synchronization with the desktop
2) more intuitive interface
3) easy access control beyond nciphub community
4) as I've heard, GoogleDrive can be integrated in some way with nciphub, so in a way the answer could be - "sure, it is already on nciphub" without any extra work
5) consistent with our use of GoogleDrive for the attributes spreadsheet - I guess we could use nciphub databases (if this is the right name) for that purpose as well, but I do not see the added benefit, and the process of editing is rather cumbersome
By the way, are there any existing examples of utilizing nciphub for a similar purpose?
Report abuse
Jayashree Kalpathy-Cramer @ on
@Andrey:
yes, the PET segmentation challenge (phase 1 and 2) used NCIP hub to share phantom data.
Essentially create a project within a group and go to files and upload/download. And sync with google drive.
Report abuse
Andrey Fedorov @ on
Sounds good, so how about we just use google drive folder I created, and then you just sync it back into nciphub?
I am not familiar with that process, so your expertise would be helpful.
Report abuse
Dariya Malyarenko @ on
We have already attempted to include these sources (wiki and converters) into the summary that we have.
In lines with QIN ADC challenge, this discussion focuses on DWI (*not* DTI), so DTI attributes are omitted on purpose.
>> If you detect missing DWI attributes, please edit/add as needed (by following the link to Google docs), as long as you have a DWI-DICOM data example to support your entry.
>>Current spreadsheet has the list of attributes for which we have DWI-DICOM examples (to be uploaded soon).
>>This library is just the "seed", and we hope and invite more data to be added by all sites that have different MRI systems and SW versions.
Report abuse
Dariya Malyarenko @ on
Note that the same references/links are listed on group's wiki-page: https://nciphub.org/groups/qindicom/wiki/DiffusionrelatedDICOMtags:experienceacrosssites
we will add the "data source" links when the examples are uploaded
Report abuse
Dariya Malyarenko @ on
We included standard tags for all vendors in the spreadsheet -- they are mostly "NAs"... Philips seems the closest to standard DICOM for "critical" DWI info (although some tags are still missing, as you can see from examples).
Our hope is that QIN BIDS group will help us with vendor-agnostic DWI conversion to *standard* DICOM by mid-summer. For their reference, we are creating this DICOM tag library across vendors. This will also be related back to vendors through QIN and RSNA.
Report abuse
Andrey Fedorov @ on
Dariya, to clarify, we discussed providing tools to save ADC maps in a DICOM-compliant format. Producing a tool to harmonize all these vendors into a standard-compliant shape is beyond what we can accomplish in QIICR. We can discuss this later, when we are better prepared with ADC tools.
No matter what, the work of putting together this spreadsheet and developing a collection of sample data will be a very valuable resource.
Report abuse
Dariya Malyarenko @ on
Yes, we will start with ADC maps (relying on DWI-DICOM)
DWI DICOM examples for the "seed library" are uploaded to "google-drive". The links were updated in the spreadsheet "Reference" tab.
Report abuse