[Insight-users] Re: Submission of a filter not using ImageToImageFilter as base class

[Luis Ibanez]

Hi Sonali

It is not a requirement that the submissions to
the Insight Journal must read an image and write
an image.

The fundamental point is that the submissions

      *MUST demonstrate reproducibility*.

That is, for example, if your contribution takes
as input two images and one XML file, in order to
compute as output a histogram and a vector. What
we want to verify is that if we compile and run
your source code, passing to it your same input,
we will obtain the same output histogram and vector.

Applications can be submitted to the Journal, we
have already received several applications. The
point again is that your submission must have
instructions on how any reader could *REPEAT*
exactly what you are reporting.

The ideal mechanism could be something like a
step by step list of instructions saying:

1) start the application
2) Load input image X1
3) Load input image X2
4) In the menu XX, set the parameter YY to 3.5
5) Select the checkbox TTT to the value KK
...
6) Click on the button "compute"
7) Save the histogram using the menu "file"


With such a clear list of instruction we could
expect readers (and reviewers) to *REPEAT EXACTLY*
what you have done on your side.


In some cases, it is possible to automate this
sequence of steps by providing scripts that run
the application. I don't know if your application
can be run in this way.


In you keep in mind that the final goal of the
Insight Journal is to resuscitate the practice of
verifying reproducibility, then, it should be
clear what the content of a submission might be.

In the ideal case, we would prefer to have applications
that allow to run an automatic session to verify
reproducibility, but, if that's not possible, we
will be happy to see a clear list of instructions
on how to manually re-run such session.


Please let us know if you have further questions


     Thanks


        Luis


----------------------
Sonali Barua wrote:
> Hello,
> 
> I have developed some graph classes for itk and I noticed in your
> submission criterions that the classes have to read in an image and
> output an image or rather have real world application. I will be
> submitting one application to ITK and it will use just one of the
> classes. The others however I just have small examples written that
> aren't real world. When you say real world, do you expect the
> application to be on medical data alone?
> 
> I appreciate your time in considering my queries,
> 
> Sincerely,
> 
> Sonali Barua.
>