3 min read

An overview to MRI software

There is a plethora of software available for imaging analysis. When also considering tools, libraries, and packages, the amount of possibilities are so extense that it might feel somehow intimidating. Fear not. This software-related anxiety will eventually dissapear. Meanwhile, let me break down the basics for you.

Let’s classify each software by their most representative feature: reconstruction, visualisation, preprocessing, and analysis.

Reconstruction. This is the first stone before going any further. Images came out from the scanner as independent files that you have to re-assemble together to make some sense out of them. The most used software for reconstruction are by far dcm2nii and MRIconvert.

Visualisation. Every single major imaging software counts with some type of visual display (e.g., FSL’s FSLeyes). Still, if you like to quickly eyeball your images without the pain, off you go to MRIcroN.

If you are fond to display your results with some sort of elegance, I highly recommend you to get familiar with SurfIce. In case you are a diehard fan of SPM, then the XJview will probably fancy you as well.

Preprocessing. This is, roughly, all the operations you must perform to make your images comparable against each other. I am not commenting on what is done in here because I will cover this in furher posts. What you need to know is, as the same as with visualisation, every major program has its preprocessing pipeline. Equally, depending on the software, this pipeline could be somehow flexible.

NOTE: There are some preprocessing steps that may benefit, or perform better, with supplemental software. I will account for this in due time.

Analysis. Last but not least. This is it. There are more versatile software allowing you to perform almost anything on MRI analysis. Cutting the chase, you should put all your money in either FSL, SPM, or AFNI.

Which one is best? This is a matter of taste. What you need to know is FSL and AFNI are freely distributed and both run in Unix (i.e., OSX, Linux). On the contrary, SPM runs in MATLAB, which means that any operating system would be able to run SPM. The bad news is (sad violin music starts) MATLAB is not free.

NOTE: There is no an “until death does us apart” in here. You can combine the aforesaid software to overcome each program’s particular pitfalls (e.g., FSL and skull-stripping; don’t panic, we will get there).

Other specialised software are Freesurfer (cortical thickness), MRItrix3 (diffusion and tractography) or the CONN toolbox (functional connectivity). There are, literally, tons of specific-programs out there, but I feel like covering only those I am most familiar with by now.