What would you like to see added in fMRIPost-Phase?
Issue #3 flags complex-valued GLM and spatial-source-phase ICA as possibly out of scope, and downstream modeling probably does belong elsewhere. However, the δφ time series itself is just a derivative, not a modeling choice — computing it via complex division relative to a reference frame is straightforward and lets downstream users do whatever modeling they prefer (standard GLM on phase, complex-valued analyses externally, phase-only connectivity) without fMRIPost-Phase taking a position.
Proposed output:
<source_entities>[_space-<label>]_desc-deltaPhase_bold.nii.gz — temporal phase changes relative to a reference timepoint, extracted via complex division, in radians
Reference frame choice (middle volume vs. mean vs. median) probably warrants a CLI flag.
- Chen, Z., Calhoun, V. (2018). Effect of objective dynamic information on the phase information in functional MRI BOLD signal. (complex-division extraction of δφ)
- Chen, J.E. et al. (2019). Resting-state "physiological networks". https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6431653/ (δφ-based RS-FC showing different sparseness/balance properties than magnitude FC)
Smaller-scope alternative to fully implementing #3; the two aren't mutually exclusive.
Do you have any interest in helping implement the feature?
Yes
Additional information / screenshots
Identified from a lit search by Claude.
What would you like to see added in fMRIPost-Phase?
Issue #3 flags complex-valued GLM and spatial-source-phase ICA as possibly out of scope, and downstream modeling probably does belong elsewhere. However, the δφ time series itself is just a derivative, not a modeling choice — computing it via complex division relative to a reference frame is straightforward and lets downstream users do whatever modeling they prefer (standard GLM on phase, complex-valued analyses externally, phase-only connectivity) without fMRIPost-Phase taking a position.
Proposed output:
<source_entities>[_space-<label>]_desc-deltaPhase_bold.nii.gz— temporal phase changes relative to a reference timepoint, extracted via complex division, in radiansReference frame choice (middle volume vs. mean vs. median) probably warrants a CLI flag.
Smaller-scope alternative to fully implementing #3; the two aren't mutually exclusive.
Do you have any interest in helping implement the feature?
Yes
Additional information / screenshots
Identified from a lit search by Claude.