Alternative labeling tool: a minimal algorithm for denoising single-subject resting-state fMRI data with ICA-MELODIC

Area of Science:

• Neuroimaging
• Computational Neuroscience

Background:

• Subject-level Independent Component Analysis (ICA) is crucial for denoising resting-state functional MRI (fMRI) data.
• Existing methods like ICA-FIX and ICA-AROMA are effective but computationally intensive and require complex setups.

Purpose of the Study:

• Introduce Alternative Labeling Tool (ALT), a user-friendly and computationally lightweight toolbox for labeling independent signal and noise components in fMRI data.
• To provide an accessible alternative for ICA-based denoising in resting-state fMRI.

Main Methods:

• ALT utilizes two manually tunable features: the proportion of an independent component's spatial map within gray matter and the positive skew of its power spectrum.
• The toolbox is integrated with FMRIB's Statistical Library (FSL).
• Validation was performed using the Open Access Series of Imaging Studies (OASIS-3) ageing dataset (n=275).

Main Results:

• ALT demonstrated a high degree of inter-rater agreement with manual labeling, achieving over 86% true positives for both signal and noise components on average.
• The tool proved effective on a dataset of 275 participants.

Conclusions:

• ALT is a viable and extensible solution for ICA-based denoising of resting-state fMRI data, suitable for both small and large-scale studies.
• This tool can facilitate wider adoption of ICA-based denoising techniques in neuroimaging research.