This is a conceptual question about qsiprep’s reconstruction pipelines: Reconstruction — qsiprep version documentation
Several of the workflows indicate that they produce both probabilistic and deterministic reconstructions. Coming from a background of using probtrackx in FSL, I am puzzled by this claim.
- Does it mean that the outputs cannot be clearly classified as deterministic vs probabilistic?
- Does it mean that some of the outputs are deterministic but other outputs are probabilistic?
Any insights into how to interpret this claim would be appreciated.
Hi @Dianne_Patterson, the mrtrix tractography (iFOD2) creates probabilistic streamlines, unlike probtrackx which produces images.
Thank you for your answer. I do understand the difference between images and streamlines!
I am wondering, however, about workflows like pyafq_tractometry, which claim to do both probabilistic AND deterministic tractography. In what sense is the tractography BOTH?
Ah! I’m sorry I misunderstood your question. It says “Both” in the documentation because you can configure the workflow to use either. PyAFQ can take probabilistic streamlines from mrtrix or it can generate its own deterministic streamlines using DIPY. The DIPY workflows can produce FOD images that can be sent to mrtrix for probabilistic tractography or to dsi studio for deterministic tractography.
The documentation could use some clarification here because these pipelines don’t produce both deterministic and probabilistic streamlines by default.
Thank you! That does help.