# Spacing between slices in NM data

• 10-02-2007, 10:20 AM
unix
Spacing between slices in NM data
I am currently trying to determine the correct mapping for NM DICOM
data generation. The problem that I have is in determining how the
Spacing Between Slices should be set (i.e. positive or negative) when
the images have been acquired feet first or head first. The definition
of the spacing between slices in DICOM part 3 is somewhat comfusing in
that it uses the terms 'behind' and 'in front' to determine the sign of
the spacing.

Any ideas?

• 10-02-2007, 10:20 AM
unix
Re: Spacing between slices in NM data
Kristin,
I asume your concern is more about how to display (build a 3D volume) from
such data.
The Standard defines Image Position ( 0020,0032) and Image Orientation
(0020,0037) in NM Detector Module (Part 3 C.8.4.11), which provide the exact
position of the first slice (frame) in the Patient Coordinate System. It
also defines Spacing Between Slices (0018,0088), in NM Reconstruction Module
(Part 3, C.8.4.15), which defines the position of the following slices
(frames) in the pixel element. Each subsequent (frame) slice from the pixel
element is parallel to the first and positioned along a normal to the first
slice. The normal (vector) is defined as a cross product of the row vector
with the column vector of the image slice. If Spacing Between Slices
(0018,0088) is positive, then the second slice will be positioned behind the
first. Now I agree that there is point of confusion here. The confusion
comes from how we interpret the word "behind". I (and I hope others)
interpret it as
"behind the previous slice along the normal vector". If Spacing Between
Slices (0018,0088) is negative the second slice is in front of the first
(i.e. in exactly oposite direction to the normal vector). Since the Standard
adopts a right-handed Cartesian coordinate system this would mean that for
example the following would apply:
===First sllice in trans-axial orienation with Image Orientation=(1,0,0;
0,1,0), and Spacing Between Slices (0018,0088)= +1, will place the second
slice behind the first, which would mean on 1mm from the first along the +z
axis, which coicides with the direction of the normal to the image plane.
===First sllice in trans-axial orienation with Image Orientation=(-1,0,0;
0,1,0), and Spacing Between Slices (0018,0088)= +1, will place the second
slice behind the first, which would mean on 1mm from the first along the -z
axis, wich this time coincides with the direction of the -z axis.
===First sllice in trans-axial orienation with Image Orientation=(1,0,0;
0,1,0), and Spacing Between Slices (0018,0088)= -1, will place the second
slice in front of the first, which would mean on 1mm from the first along
the -z axis
===First sllice in trans-axial orienation with Image Orientation=(-1,0,0;
0,1,0), and Spacing Between Slices (0018,0088)= -1, will place the second
slice behind the first, which would mean on 1mm from the first along the +z
axis. etc.
similarily
===First sllice in coronal orienation with Image Orientation=(1,0,0;
0,0,-1), and Spacing Between Slices (0018,0088)= +1, will place the second
slice behind the first, which would mean on 1mm from the first along the +y
axis
===First sllice in coronal orienation with Image Orientation=(-1,0,0;
0,0,-1), and Spacing Between Slices (0018,0088)= +1, will place the second
slice behind the first, which would mean on 1mm from the first along the -y
axis
===First sllice in coronal orienation with Image Orientation=(1,0,0;
0,0,-1), and Spacing Between Slices (0018,0088)= -1, will place the second
slice in front of the first, which would mean on 1mm from the first along
the -y axis
===First sllice in coronal orienation with Image Orientation=(-1,0,0;
0,0,-1), and Spacing Between Slices (0018,0088)= -1, will place the second
slice behind the first, which would mean on 1mm from the first along the +y
axis etc.
and
===First sllice in sagittal orienation with Image Orientation= (0,1,0;
0,0,-1), and Spacing Between Slices (0018,0088)= +1, will place the second
slice behind the first, which would mean on 1mm from the first along the -x
axis
===First sllice in sagital orienation with Image Orientation= (0,-1,0;
0,0,-1), and Spacing Between Slices (0018,0088)= +1, will place the second
slice behind the first, which would mean on 1mm from the first along the +x
axis
===First sllice in sagittal orienation with Image Orientation= (0,1,0;
0,0,-1), and Spacing Between Slices (0018,0088)= -1, will place the second
slice in front of the first, which would mean on 1mm from the first along
the +x axis
===First sllice in saggital orienation with Image Orientation= (0,-1,0;
0,0,-1), and Spacing Between Slices (0018,0088)= -1, will place the second
slice behind the first, which would mean on 1mm from the first along the -x
axis

I hope this is true and helpful but I will ask some of the 3D gurus to
step in and
confirm.

Best regards,
Todd

>I am currently trying to determine the correct mapping for NM DICOM
> data generation. The problem that I have is in determining how the
> Spacing Between Slices should be set (i.e. positive or negative) when
> the images have been acquired feet first or head first. The definition
> of the spacing between slices in DICOM part 3 is somewhat comfusing in
> that it uses the terms 'behind' and 'in front' to determine the sign of
> the spacing.
>
> Any ideas?
>[/color]

• 10-02-2007, 10:20 AM
unix
Re: Spacing between slices in NM data
>>>>> "T" == T Kantchev <todd@kantchev.co.uk> writes:
[color=blue]
> [...] defines Spacing Between Slices (0018,0088), in NM
> Reconstruction Module (Part 3, C.8.4.15), which defines the
> position of the following slices (frames) in the pixel
> element. Each subsequent (frame) slice from the pixel
> element is parallel to the first and positioned along a
> normal to the first slice. The normal (vector) is defined
> as a cross product of the row vector with the column vector
> of the image slice. If Spacing Between Slices (0018,0088)
> is positive, then the second slice will be positioned
> behind the first. Now I agree that there is point of
> confusion here. The confusion comes from how we interpret
> the word "behind". I (and I hope others) interpret it as
> "behind the previous slice along the normal vector".[/color]

I believe your confusion can be removed if you think in terms of
geometry and ignore the "in front" and "behind" parts. If the
vector position of the first slice is r1, the unit vector normal
to the first slice is n, and the slice spacing is s, then the
positon of each slice in space is r1 + s n. The slice order
relative to the direction of n is taken care of by the sign of
the slice spacing s.

Mike
• 10-02-2007, 10:20 AM
unix
Re: Spacing between slices in NM data
Hi

OK the problem that I am still not sure about is this. If I have a
head first scan and my first image taken is of the head then the last
image will be of the feet. In a Head First Supine scan the Image
Orientation (Patient) will be (1,0,0,0,1,0) and the normal will be
(0,0,1).

In the above situation am I correct in setting the spacing between
slices negative thus ensuring that the first slice of the head has the
maximum z value, while the last slice (of the feet) has the minimum z
value? I hope that I have at least understood this orientation
correctly.

However for a Feet First Supine scan the Image Orientation (Patient)
will be
(-1,0,0,0,1,0) and the normal will be (0,0,-1). If the first image
taken is that of the feet and the last image taken is that of the head
what should the spacing between slices be? Should the spacing between
slices be positive or negative? The thing that is confusing me is that
the normal is in the direction (0,0,-1) however the patients Z axis is

[color=blue]
>
> Mike[/color]

• 10-02-2007, 10:20 AM
unix
Re: Spacing between slices in NM data
Kristin,
I think you have to be clear on which side of the imaging you are. If you
are displaying, then it does not matter how the patient has been scanned.
What matters is the position and the orientation of the first slice (frame)
from the sequence of frames in the pixel element. Each subsequent frame from
the pixel element has a position which can be calculated by Michael's
formulae and the orientation is parallel to the first one.
However, if you are on the acquisition side it would depend on the
reconstruction algorithm, rendering the images and also on the order in
which you place the reconstructed frames in the pixel element. Let us assume
that the order of the frames in the pixel element is the same as the order
in which the slices are acquired as the patient moves towards the gantry
(bare in mind that some scanners may not have such technique of scanning,
because they may use a single cone beam, that takes a whole body scan on one
rotation and the slicing is produced without bed movement. There are also 3D
whole body capturing techniques in PET that also are very confusing in
trying to understand when the head is first and when the feet are first). If
we think in the category of bed-gantry movement and if the reconstruction
renders the images as if they are looked at from the bed towards the gantry,
then your first case (head first) is correct. For the second (feet first) I
think you would be right to set spacing between slices positive.
Best regards,
Todd
> Hi
>
> OK the problem that I am still not sure about is this. If I have a
> head first scan and my first image taken is of the head then the last
> image will be of the feet. In a Head First Supine scan the Image
> Orientation (Patient) will be (1,0,0,0,1,0) and the normal will be
> (0,0,1).
>
> In the above situation am I correct in setting the spacing between
> slices negative thus ensuring that the first slice of the head has the
> maximum z value, while the last slice (of the feet) has the minimum z
> value? I hope that I have at least understood this orientation
> correctly.
>
> However for a Feet First Supine scan the Image Orientation (Patient)
> will be
> (-1,0,0,0,1,0) and the normal will be (0,0,-1). If the first image
> taken is that of the feet and the last image taken is that of the head
> what should the spacing between slices be? Should the spacing between
> slices be positive or negative? The thing that is confusing me is that
> the normal is in the direction (0,0,-1) however the patients Z axis is
> (0,0,1) from feet to head.
>
>
>
>
>
>[color=green]
>>
>> Mike[/color]
>[/color]

• 10-02-2007, 10:20 AM
unix
Re: Spacing between slices in NM data
P.S. You have to be also mindful of Slice Vector as well, which may change
the slice order. In the above examples we assume it progresses from low
frame numbers towards high. And of course we are only talking about RECON
TOMO and RECON GATED TOMO. Other Image Types are not volumetric and do not
have exact location in space. Complicate matter that NM... :-)

> Hi
>
> OK the problem that I am still not sure about is this. If I have a
> head first scan and my first image taken is of the head then the last
> image will be of the feet. In a Head First Supine scan the Image
> Orientation (Patient) will be (1,0,0,0,1,0) and the normal will be
> (0,0,1).
>
> In the above situation am I correct in setting the spacing between
> slices negative thus ensuring that the first slice of the head has the
> maximum z value, while the last slice (of the feet) has the minimum z
> value? I hope that I have at least understood this orientation
> correctly.
>
> However for a Feet First Supine scan the Image Orientation (Patient)
> will be
> (-1,0,0,0,1,0) and the normal will be (0,0,-1). If the first image
> taken is that of the feet and the last image taken is that of the head
> what should the spacing between slices be? Should the spacing between
> slices be positive or negative? The thing that is confusing me is that
> the normal is in the direction (0,0,-1) however the patients Z axis is
> (0,0,1) from feet to head.
>
>
>
>
>
>[color=green]
>>
>> Mike[/color]
>[/color]