odlgroup/odlemrecon

ODL bindings for EMrecon

odlemrecon's Issues

Infer domain and range from settings

Currently the domain and range are magically computed from the settings, we need to be able to infer these so that users don't have to give magic parameters.

sh: 1: EMrecon_siemens_pet_tools: not found

Hello, I am trying to perform the example code "mlem", but it cannot find the EMrecon_siemens_pet_tools.
From the previous issues, I see that authors "have hard-coded EMrecon_siemens_pet_tools", I wonder if there is something wrong with my settings.

Reuse files

The temporary files are currently re-created in each iteration. If they were instead re-used we could see a performance improvement.

Document what is computed

The values computed are currently quite vacuous with no units nor further explanations. We need to either refer to the correct documentation or do this ourselves.

Fix scaling of operators

The operators currently operate on "matrices", we need to fix the scaling so that they property compute the correct thing regardless of sizes.

Support non-siemens pet

Currently we have hard-coded EMrecon_siemens_pet_tools this should be selectable by users

Add attenuation correction

The attenuation can be computed by simply forward projecting the mu-map and taking the negative exponential of it (given that we have the correct units, see #1 and #2). We can use this to make attenuation corrected reconstructions by simply left multiplying with this correction factor.

Add example with sensitivities

We currently don't use the sensitivities in the examples, we need an example of using this.

Add MLEM example

MLEM is the standard reconstruction scheme in emission tomography, we need an example of it.