Yes, you understand.

The trigger position can also be negative and therefore occur before the
multiplexed group. It is not necessary to offset the skew of the waveforms
to achieve this.

The trigger position has no automatic place within the samples. Its primary
purpose is to synchronize this multiplex group with some other event that
occurs outside this DICOM object but could be found in another DICOM object
or be relevant to the context.

The term Trigger is a generic label that is understood within the context of
the procedure. It could be a camera shutter, it could be an injection bolus,
it could be an audio stimulus. It is something generally not captured on one
of the channels as a trigger waveform (though if the equipment was so
capable it could have been).

The original intent of skew was to define the minute time between
multiplexed channels not acquired absolutely simultaneously as with a
latching sampler but uses a single A/D converter sequentially rotating
through the channels. Skew in this case is generally quoted in time and
expected to be less than the sampling interval.

However, there are other instances when larger delays occur such as data
picked up after a long delay line of digital filtering applied to one
channel but not others. Skew in this case is generally quoted in samples.


"Michael Onken" wrote in message
> Hi Richard,
> Richard Crane wrote:
> > [general explanations]

> Thanks:-)
> > In your example there is a skew = 7 between channel #1 and channel #2 so
> > if the Trigger occurs at sample 31 (relative to the multiplexed group)
> > then it will occur after the 30 samples of channel #1 and at sample 24

> > channel 2.

> So you just count the samples as they occur in the Waveform Data Tag

> e.g: C1S1C2S1C3S1 C1S2C2S2C3S2 C1S3C2S3C3S3 (spaces for readability).
> Last,third "group" is then Trigger Sample Position 3. This seems to be

> in the supp 30 when it says: "...all channels are considered to be
> synchronized". Hope I got this right now!
> > The first sample of Channel #2 will be the second sample of the
> > multiplexed group - (it does not follow on after channel #1) but when it
> > appears on a display then it will be superimposed under sample 7 of
> > channel #1, and above sample 5 of channel #3.

> This is the way I drew it, I think, ok:-)
> In my example channel #1 has no sample skew. Since this is not required, I
> must ask another question to get this right for me. As I understand, the
> Trigger Sample Pos doesn't define a sample, that occurs at the same time

> the Trigger, but defines a time relative to the multiplex group start

> measured in samples.
> The Trigger Sample Position is always relative to the Multiplex Group
> Start Time. This would mean, that I could calculate the time, at which the
> Trigger occurs, without even knowing the samples and their channel skews /
> offsets - just using the sampling frequency and the sample "count"

> Sample Pos). When displaying the channels, I would draw them based on the
> sample skew and latency and after all i draw a marker at the time, where
> the trigger occurs for example.
> This would give me the possibility to define a Trigger Sample Pos defining

> time, that isn't in the "time window" of the samples of the channels
> itself, if they are arranged considering their sample skew. (E.g. all
> channels have sample skews 100,110,120,.. and the Trigger Sample Pos is

> to 80).
> In case of a of a Trigger Signal to a camera, a short delay between

> and first photo shot would make even sense ...seems all a bit mysterious

> me, anyway.
> All corrections are welcome:-)
> Thank you very much
> Michael