Read input images from real-time dicom export

[ohinds]

This adds the functionality to read input images from a folder that's the target of a real-time dicom export from the scanner. It's designed to support the Siemens BOLD add-in.

I've also updated the regression test and example data to include a dicom test.

[pwighton]

Awesome!! I'll try to find some scanner time next week to test this out.

[satra]

perhaps install this from their github release directly. see the instructions in this binaries section: https://github.com/ReproNim/neurodocker/blob/24d84b149b02e349de49398438299f9a2a712dec/neurodocker/templates/dcm2niix.yaml#L13

[ohinds]

is there specific reason we'd think the apt-gettable version isn't sufficient? i'd rather not increase the build time unless we have to.

[satra]

it's not always up to date. you can check how many revisions it lags by.

[satra]

also this shouldn't add any build time. did you see the instructions in the binaries section?

[ohinds]

ah gotcha. i guess my remaining question is whether there are specific concerns with a lagged version.

[satra]

i think for purpose of this PR less of a concern. this is a non-trivial conversion. i would worry if someone tries to use multi-echo data and more generally if this works well across scanners. for purpose of your target group this is less of a concern.

[ohinds]

that makes sense. i'll make an issue.

[satra]

only skimmed the code, but looked reasonable.

[pwighton]

Added a couple suggestions to help us understand what's going on when dcm2niix fails and has to retry. Otherwise, code looks good. Will hold off on approval until I've confirmed it's working on the scanner.

[ohinds]

@satra @pwighton a note for y'all. to take advantage of features in newer g++, i bumped the base image here to 24.04. i don't think that should be a huge issue but if you wanted to take a look at that last commit again that might make sense.

[pwighton]

Cool. That shouldn't be an issue, but will test it tonight.

[pwighton]

Confirmed working on xa30.