Problems in reading Hounsfield units from a DICOM image (Fiji)

[Shreyakanakiya]

Hello,

I have a set of DICOM images. When I import them as an image sequence I believe that Fiji reads the slope and intercept from the DICOM header and converts the raw pixel values to Hounsfield units (HU). In my header file, however, the rescale slope = 1 and rescale intercept = 0. Thus by the formula,

Hounsfield unit = pixel value*slope+intercept

My Hounsfield units are the same as pixel values. I want to segment the image using the Hounsfield scale (air = -1000 HU, water = 0 HU). But since all my values are positive I am unable to use it directly. I have attached a histogram of my DICOM image stack for reference.  The left peak represents the background (air) and the right peak represents the grayscale distribution of my data (rock).

I'm not aware of a calibration to have only positive HU but am new to this field and wanted to know if such a calibration exists ? If yes, is there a way I can rescale them such that air = -1000 HU, water = 0 HU and so on?  

Thanks in advance.

-Shreya

[Salim Kanoun]

Hi Shreya,

I never seen that.
Depending on the manufacturer there is a high range of possible rescale
slope/intercept.
However imageJ always recover the correct UH for each CT I have seen.

Maybe your Dicom are corrupted ? Can you share with us a sample of the
original dicom file ?

Best regards

Salim



Le 26 févr. 2017 06:24, "Shreyakanakiya" <[hidden email]> a écrit :

Hello,

I have a set of DICOM images. When I import them as an image sequence I
believe that Fiji reads the slope and intercept from the DICOM header and
converts the raw pixel values to Hounsfield units (HU). In my header file,
however, the rescale slope = 1 and rescale intercept = 0. Thus by the
formula,

Hounsfield unit = pixel value*slope+intercept

My Hounsfield units are the same as pixel values. I want to segment the
image using the Hounsfield scale (air = -1000 HU, water = 0 HU). But since
all my values are positive I am unable to use it directly. I have attached a
histogram of my DICOM image stack for reference.  The left peak represents
the background (air) and the right peak represents the grayscale
distribution of my data (rock).

I'm not aware of a calibration to have only positive HU but am new to this
field and wanted to know if such a calibration exists ? If yes, is there a
way I can rescale them such that air = -1000 HU, water = 0 HU and so on?

Thanks in advance.

-Shreya



<http://imagej.1557.x6.nabble.com/file/n5018195/histogram.png>



--
View this message in context: http://imagej.1557.x6.nabble.
com/Problems-in-reading-Hounsfield-units-from-a-DICOM-
image-Fiji-tp5018195.html
Sent from the ImageJ mailing list archive at Nabble.com.

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Shreyakanakiya]

Hi Salim,

I have attached a sample of the DICOM file here.
Thanks!

-Shreya

Image.dcm

[Salim Kanoun]

Dear Shreya,

Your dicom header is almost empty.
For example there is no "modality" tag, that should say that the
modality is CT. Usually we also find a lot of other informations like
the manifacturer name, the XRay kV used ect ect...

I don't think the problem is in Fiji.
Fiji is showing you all you can get from the file and obviously the
unit don't seems to be UH.

I think you should see the manufacturer documentation of your CT device.

I think there are two possibilities :
1) The unit are not UH and your CT might use another scale, in that
case probably that your manufacturer could give you a conversion rule
to get UH.
2) The original might been transformed by another software and some
tag have been deleted or changed leading to a wrong scale. Are you
sure that this file is the original output file from your CT device ?


Best regards,

Salim


2017-02-26 12:49 UTC+01:00, Shreyakanakiya <[hidden email]>:

> Hi Salim,
>
> I have attached a sample of the DICOM file here.
> Thanks!
>
> -Shreya
>
> Image.dcm <http://imagej.1557.x6.nabble.com/file/n5018198/Image.dcm>
>
>
>
> --
> View this message in context:
> http://imagej.1557.x6.nabble.com/Problems-in-reading-Hounsfield-units-from-a-DICOM-image-Fiji-tp5018195p5018198.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html