STORM-NET

STORM-NET estimates fNIRS optode positions from video using computer-vision to enable accurate optode-to-scalp co-registration for spatially precise fNIRS analysis in adults and developmental populations such as infants.

Key Features:

• Video-based appearance model: Employs a fully-automatic appearance-based computer-vision method to estimate registration parameters of a mounted cap to the scalp from raw video.
• Speed and accuracy: Produces spatial accuracy comparable to existing approaches while operating orders of magnitude faster.
• Motion-tolerant registration: Operates with moving subjects and has been demonstrated usable with infants and adult subjects.
• Validation against standards: Validated with comparative analyses against standard 3D digitizers and other photogrammetry-based approaches.
• Applicability in uncontrolled settings: Can be applied in non-laboratory or unconventional environments for fNIRS co-registration.

Scientific Applications:

• Optode co-registration: Enables accurate co-registration of optode positions to the scalp to improve spatial localization in fNIRS studies.
• Developmental neuroscience: Facilitates fNIRS investigations in infants and other developmental populations where motion complicates traditional registration methods.
• Naturalistic and longitudinal studies: Supports collection of spatially precise fNIRS data in naturalistic environments and longitudinal infant studies outside controlled laboratory settings.

Methodology:

Fully-automatic appearance-based computer-vision estimation of cap-to-scalp registration parameters from video, with validation on adult subjects, demonstrations with infants, and comparative analyses against 3D digitizers and photogrammetry-based approaches.