use dicom's Physical Delta X as spacing?

Hi, I’m trying to deal with the ultrasound images. And there is a Physical Delta X attribute (see here) in the DICOM files. Should it be used as spacing? I read the DICOM files with simpleitk, but the spacing is not the same value as Physical Delta X. Any idea?

Hie @hubutui,

This is a bit tricky. The Physical Delta X attribute is not necessarily geometric spacing, it can have different meanings based on the image content (it can be in time for Doppler or geometric space for anatomy).

You will probably also have to check Physical Units X Direction. Note that in ITK/SimpleITK all physical units are in mm, so if this specifies the units in cm they will be converted to mm when read.

Yeah, I have check Physical Units X Direction too. Luckily, in my dataset, it’s only related to geometric space for anatomy. And they have the same value in Physical Delta X and Physical Delta Y.
I wanna calculate the hausdorff distance in a segmentation task. I think simpleitk could help with be aware of the physical unit.

Sometimes it works, if there is one big tissue region. That approach exists is some scripts, but i wouldn’t recommend generally. Attributes are from US Region Calibration Module,
there are often many regions and they can overlap, delta value can be zero or negative.
In fact it is not spacing, it is for measurements in particular region, not whole image.
For spacing there may be Pixel Aspect Ratio (0028,0034) attribute (some data sets may use Pixel Spacing).

Here are e.g. 3 regions, PW Spectral Doppler region (blue) has different units and deltas for X and Y directions
20190506-042607

1 Like

Oh, I see. Luckily, in my task. There is only one region. This assumen also helps to remove unexpected banner regions.

2 Likes

Good day,
i have a question
is this correct (0018,602C) PhysicalDeltaX -0.036248147487640381 according to dicom standards of US image?

can the physicalDelta X (0018,602C) have a negative value in the US image according to dicom standard?

thank you for your reply in advance

1 Like

Hello @hubutui!
I’m curious, did your project work?

I’m also calculating distance, and I also only have these values (Physical Delta) and I don’t have Pixel Spacing. So I’m trying to do the exact same thing as you.

Monica

sorry, I move on to another project. I’m not sure if this work. I think we could assume pixel spacing = physical delta when only on is available in the metadata.

Ok. Thanks for your reply! :slight_smile:

Hi @monica_heddneck,
In my case the PhysicalDeltaX is also ignored in itk’s spacing, although the unit is cm (PhysicalUnitsXDirection == 3).
I can access these attributes in python with pydicom:

  • ds.SequenceOfUltrasoundRegions[0].PhysicalDeltaX
  • ds.SequenceOfUltrasoundRegions[0].PhysicalDeltaY
  • ds.SequenceOfUltrasoundRegions[0].PhysicalUnitsXDirection
  • ds.SequenceOfUltrasoundRegions[0].PhysicalUnitsXDirection

I am not sure what is the expected behavior of itk in this case.

1 Like

It’s definitely an interesting field.
From what I’ve learned, if a patient’s kidney, heart, etc, is deeper in their torso, this value needs to be adjusted during data capture.