Message |
Date: 2010-07-13 19:15
Sender: Colin Shea
I think it's a situation where
using the action discovery tool probably isn't going to work in
this use case. The way that mipav algorithms internally handle
multiple inputs specified by the user seems to be incompatible with
the JIST interface. For instance, if I want to specify the
inversion time (TI) for the second T1IR volume there is no place to
define it since it is only expecting 1 T1IR volume as the default.
The parameter labels in the JIST module have names like
"SPGR_TRunsupportedList", so I think there is an internal list
structure to the algorithm that can handle multiple inputs
specified on the fly, but it isn't transferring over to the JIST
module through the action discovery tool. I think it would be best
to write a dedicated module from scratch that uses the JIST
conventions for setting multiple input volumes and multiple
parameters. If you need me to specify these properties in detail I
can do so.
|
Date: 2010-07-12 17:34
Sender: Bennett Landman
JIST uses exactly the
information that action discovery provides. Action =
discovery does not currently well-capture plugins that take
variable =
numbers of inputs - we allow you to specific any options at
runtime. Are =
the "mandatory" fields not set correctly within the flip angle
=
calculation?=20
|
Date: 2010-07-12 17:31
Sender: Colin Shea
I've updated to MIPAV v1.5
(the newest release) and there is now a discoverable plugin called
"FlipCalcT1(5.0.0 R)" under MIPAV->Algorithms->MRI.
Unfortunately we're running into the same problem. The MIPAV plugin
works fine when using it manually to create T1, B1, etc maps
because I can specify how many SPGR volumes and how many IR volumes
to use. The GUI will update itself to accept more parameters when
multiple volumes are selected. However, the JIST version of this
plugin doesn't know a priori how many volumes will be specified.
Therefore the input pins that should show up for the input volumes
don't exist. Furthermore, I can't specify the IR times for the two
IR volumes as well as some other pertinent parameters. I think it
is a matter of the ActionDiscovery interface not being compatible
with the way JIST defines input volumes. I think in order to
properly set up the JIST module version of this plugin, we need
volume collection input pin for SPGR volumes and another volume
collection pin for IR volumes. That way it can handle 0,1,2+
volumes for each type. Similarly we need input pins that can take
0,1,2+ parameters like flip angle and inversion time.
|
Date: 2010-03-26 19:36
Sender: Colin Shea
I got a response from Dr.
McAuliffe; apparently Justin
Senseney implemented the despot plugin at Danny's request and
they plan to implement the despot-JIST API "in the near
future".
|
Date: 2010-03-12 17:32
Sender: Bennett Landman
Do you know who developed the
original MIPAV plugin? If so, can you ask =
that person to implement the "ActionDiscovery" interface. If this
is =
implemented, the native MIPAV plugin will automatically be detected
in =
JIST. It's a very light weight interface to add to a module that
already =
supports scriptable.
|
Date: 2010-03-12 17:26
Sender: Bennett Landman
(enter your response
here)
|