See

http://www.dclunie.com/medical-image...COMOrientation

markww wrote:
> Hello all,
>
> I am trying to figure out how to read the direction the rows and
> columns of DICOM images travel using the image orientation patient tag,
> I am presenting here how I would read it. Could someone please verify
> if this is correct? Thanks.
>
> The orientation tag has 6 components, xyz orientation of both the ROWS
> *then* COLUMNS of the image. So there is this mapping for both rows and
> columns:
>
> if (CosineComponentX < 0)
> x component goes towards the RIGHT.
> else
> x component goes towards the LEFT.
>
> if (CosineComponentY < 0)
> y component goes towards the ANTERIOR.
> else
> y component goes towards the POSTERIOR.
>
> if (CosineComponentZ < 0)
> z component goes towards the FEET.
> else
> z component goes towards the HEAD.
>
> This is a sample orientation:
> 0.000000\1.000000\0.000000\0.000000\0.000000\-1.000000
>
> So the y cosines are:
> 0.000000\1.000000\0.000000
>
> The x cosines are:
> 0.000000\0.000000\-1.000000
>
> So the orientations of the rows are then:
> Towards the n/a, POSTERIOR, n/a.
>
> And the orientations of the columns are then:
> Towards the n/a, n/a, FEET.
>
> I guess you can tell the 'major' direction they are traveling based on
> which of the component values are greatest (their absolute values)?
>
> I ask because if there is a fictional orientation string like (for cols
> lets say):
> 0.000000\0.200000\-1.000000

Illegal ... some of the squares must be 1 ... these are UNIT vectors.

> Then there must be movement along another axis, but the major direction
> is towards the FEET. But the 0.2 in the y component means that columns
> are also traveling to the POSTERIOR. So major direction FEET, minor
> direction POSTERIOR.

Thank you very much David. The code you present in your webpage has
been working well. Have you been using that code yourself consistently?
I am extremely paranoid about presenting an incorrect orientation to
the user, which could have the effect of a doctor making a call on an
organ which really belongs to the other side of the patient. Is there
really any other way to interpret the orientation tags? It looks like
this is a safe method,

Thanks