I've been thinking of a way to validate the phantom. The idea would be to find the appropriate parameters for diffusion tensors/balls and preprocessing (#2) tasks in an interative manner.
We could learn these parameters by simulating the very same gradient scheme of the original data. Then, we'd compute the error and update parameters. I'll see how to integrate the parameters in an optimization framework so that we don't rely on rules of thumb.
However, this is just a thought. If you could come with something more elaborated I'll be happy to use it :)
I've been thinking of a way to validate the phantom. The idea would be to find the appropriate parameters for diffusion tensors/balls and preprocessing (#2) tasks in an interative manner.
We could learn these parameters by simulating the very same gradient scheme of the original data. Then, we'd compute the error and update parameters. I'll see how to integrate the parameters in an optimization framework so that we don't rely on rules of thumb.
However, this is just a thought. If you could come with something more elaborated I'll be happy to use it :)