SUV formula with Rescale Intercept and Slope

Joh

unread,

Sep 20, 2017, 8:15:56 AM9/20/17

to

Dear all,

I am working with PET CT and I want to display the SUV in a good manner.
But when I check different biobliographies, they are all ok about the suv factor but the application of rescale slope and intercept differs (even if the Rescale intercept should be zero).

I have the classical formula (I think)

Pixel_Value = (Pb * Rs + Ri) * SUVf [1]

And another one:

Pixel_Value = ((Pb + Ri)* Rs) * SUVf [2]

where Pb is pixel buffer value; Ri Rescale Intercept and Rs Rescale Slope.
[1] is a classical formulation of the DICOM standard to rescale pixel buffer that I have seen in conformance Statements to compute the SUV such as Philips.
[2] can be found in http://qibawiki.rsna.org/index.php/Main_Page

From this observation, I was wondering if I forgot something ? Does anyone can help me to understand this difference or help me to find a bibliography ?

Thanks.

David Clunie

unread,

Sep 20, 2017, 9:20:45 AM9/20/17

to

Hi Joh

The first formula is obviously correct.

I am an idiot and copied it wrong into the QIBA SUV documents and nobody noticed. I have fixed the QIBA documents so that they are now the same.

See the updated QIBA page at:

http://qibawiki.rsna.org/index.php/Standardized_Uptake_Value_(SUV)#SUV_Calculation

and for reference the DICOM definition:

http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.11.html#para_41443e56-dcf3-4d8e-8008-6eaed66620cc

David

Joh

unread,

Sep 20, 2017, 10:16:28 AM9/20/17

to

Hi David,

Many thanks for the quick reply and for the documentation you provide to help the community.

Johan

Mihail Isakov

unread,

Jun 21, 2018, 6:39:37 PM6/21/18

to

Hi.

Thank you very much.

BTW,
"if Corrected Image (0x0028,0x0051) contains ATTN and DECAY"
In the images i got (TCIA Collections) Corrected Image attribute contains not "DECAY", but "DECY". I hope it is the same thing.

"Collected DICOM PET/CT images of the same reference phantom from scanners from GE, Philips, Siemens"
Are the images available for download?

Thanks.

Regards,
Mikhail

Mihail Isakov

unread,

Jun 22, 2018, 12:59:49 AM6/22/18

to

FYI, another issue in the document
"start Time = Radiopharmaceutical Start Time (0x0018,0x1072) in Radiopharmaceutical Information Sequence (0x0054,0x0016)
//start Date is not explicit ... assume same as Series Date; but consider spanning midnight"

But Radiopharmaceutical Start Time is deprecated, s.

http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.8.9.2.html

"The use of this Attribute is deprecated in favor of Radiopharmaceutical Start Date Time (0018,1078)
The use of a time alone can cause confusion when the procedure spans midnight."

Thank you.

Regards,
Mikhail

P.S.

So i tried to use Radiopharmaceutical Start Date Time (if available) and it has side effect - some images (from TCIA Collections) don't work longer, sometimes radiopharmaceutical date is very different from Series Date (e.g. +6 years in the future). I guess anonymizer software tried to modify dates, but forgot Radiopharmaceutical Start Date Time(0018,1078)... Shall do a workaround option.. OMG :)
Thanks.

David Clunie

unread,

Jun 26, 2018, 9:18:42 AM6/26/18

to

On Thursday, June 21, 2018 at 6:39:37 PM UTC-4, Mihail Isakov wrote:

> "if Corrected Image (0x0028,0x0051) contains ATTN and DECAY"
> In the images i got (TCIA Collections) Corrected Image attribute contains not "DECAY", but "DECY". I hope it is the same thing.

It should be "DECY" not "DECAY" per the standard; I will fix that

> "Collected DICOM PET/CT images of the same reference phantom from scanners from GE, Philips, Siemens"
> Are the images available for download?

Not AFAIK, but I will look into it.

David

David Clunie

unread,

Jun 26, 2018, 9:21:35 AM6/26/18

to

On Friday, June 22, 2018 at 12:59:49 AM UTC-4, Mihail Isakov wrote:

> "start Time = Radiopharmaceutical Start Time (0x0018,0x1072) in Radiopharmaceutical Information Sequence (0x0054,0x0016)

> But Radiopharmaceutical Start Time is deprecated

Good point. I have fixed that too. See the updated documents on the site

> So i tried to use Radiopharmaceutical Start Date Time (if available) and it has side effect - some images (from TCIA Collections) don't work longer, sometimes radiopharmaceutical date is very different from Series Date (e.g. +6 years in the future). I guess anonymizer software tried to modify dates, but forgot Radiopharmaceutical Start Date Time(0018,1078)...

That's worrying ... can you email me with details of which collections/studies/series in TCIA are affected by this and I will get on to them to investigate this and fix it and/or add a note to their documentation about it if it turns out not to be a significant privacy issue.

David (dcl...@dclunie.com)