detrending

Detrending Model Doc

Overview

Detrending is a classical non-deep-learning method for neuroimaging signal denoising.

• Model family: non-deep-learning preprocessing and denoising method
• Typical objectives:
  • remove low-frequency drift and temporal trends
  • stabilize time series before connectivity, decoding, or statistical analysis
  • prepare cleaner voxel-wise or ROI-wise time series for downstream workflows
• Primary input: preprocessed fMRI time series, optional confounds, optional mask, TR
• Primary output: cleaned BOLD image, cleaned ROI time series, optional QC summaries

In NeuroClaw, this document is model-level guidance for detrending workflows rather than predictive modeling.

Upstream preparation should usually be delegated to:

• fmri-skill for modality-level denoising planning and validated preprocessing sequences
• nilearn-tool for concrete detrending and cleaned time series export

Research use only.

---

Quick Start

1) Prepare denoising inputs

Expected inputs:

• preprocessed BOLD image
• repetition time (TR)
• optional confounds TSV
• optional brain mask

If images are not preprocessed yet, delegate to fmri-skill first.

2) Detrending route

Representative operations:

• load preprocessed image or extracted ROI time series
• remove constant and linear temporal trends
• optionally combine detrending with confound regression or standardization
• export cleaned image or time series table

Example execution route:

# delegated through claw-shell after preprocessing is confirmed
python skills/nilearn-tool/scripts/denoise_timeseries_reference.py \
  --bold path/to/sub-001_rest_preproc_bold.nii.gz \
  --confounds path/to/sub-001_confounds.tsv \
  --tr 2.0 \
  --detrend \
  --output-dir run_models_output/detrending

---

Input / Output Contract

Required inputs

• preprocessed BOLD image or extracted time series
• TR when combined with temporal cleaning workflow metadata

Optional inputs

• confounds table
• mask image
• standardization options