CortexLume: Rapid Geometry-Adaptive Cortical Feedback for Target-Informed fNIRS Design
Abstract
Target-informed functional near-infrared spectroscopy (fNIRS) probe design requires repeated evaluation of how candidate source-detector geometries sample the cortex, but full photon-transport modeling is too costly for interactive revision. We developed CortexLume, which couples an editable target-patch-placement representation with a geometry-adaptive cortical surrogate for rapid spatial assessment. The surrogate adapts its center, orientation, and anisotropic extent to fitted source-detector geometry and local scalp-cortex relationships, and is intended to approximate normalized cortical spatial allocation rather than absolute optical sensitivity. We benchmarked the method against MCX photon transport assembled through AtlasViewer on the ICBM152 anatomical model and then evaluated the frozen formulation on the independent Colin27 model without retuning. On ICBM152, mean channel-centroid discrepancy was 2.35 mm, 90%-mass Dice was 0.746 per channel and 0.826 per patch, and Jensen-Shannon distance was 0.330; the complete formulation improved Dice in 370 of 396 channels relative to a restricted reference. On Colin27, mean channel Dice was 0.748 and patch Dice was 0.812 across 88 retained channels. Patch-field construction required a median of only 75.81 ms. These results support CortexLume as a rapid cortical surrogate for iterative probe drafting and candidate shortlisting, with subject-specific photon modeling retained for final optical refinement.
Related
- Cell type specific astrocytic feedback regulates excitation inhibition balance and cortical network dynamics
- Direct cortical feedforward and feedback pathways differentially shape visual responses and selectively engage looped circuits
- Role of leg campaniform sensilla proprioceptive feedback in Drosophila melanogaster adaptive walking
- Multilingual Brain Decoding Reveals Shared Cortical Semantic Geometry
- Asymmetric cortical mechanics shape epithelial cell geometry in organoids
BibTeX
@misc{huang2026cortexlume,
title = {{CortexLume: Rapid Geometry-Adaptive Cortical Feedback for Target-Informed fNIRS Design}},
author = {Ruiyi Huang},
howpublished = {PsyArXiv},
year = {2026},
month = oct,
doi = {10.31234/osf.io/nx5st_v1},
url = {https://osf.io/preprints/psyarxiv/nx5st_v1/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - CortexLume: Rapid Geometry-Adaptive Cortical Feedback for Target-Informed fNIRS Design
AU - Ruiyi Huang
PY - 2026
DA - 2026/10/09
PB - PsyArXiv
DO - 10.31234/osf.io/nx5st_v1
UR - https://osf.io/preprints/psyarxiv/nx5st_v1/
AB - Target-informed functional near-infrared spectroscopy (fNIRS) probe design requires repeated evaluation of how candidate source-detector geometries sample the cortex, but full photon-transport modeling is too costly for interactive revision. We developed CortexLume, which couples an editable target-patch-placement representation with a geometry-adaptive cortical surrogate for rapid spatial assessment. The surrogate adapts its center, orientation, and anisotropic extent to fitted source-detector geometry and local scalp-cortex relationships, and is intended to approximate normalized cortical spatial allocation rather than absolute optical sensitivity. We benchmarked the method against MCX photon transport assembled through AtlasViewer on the ICBM152 anatomical model and then evaluated the frozen formulation on the independent Colin27 model without retuning. On ICBM152, mean channel-centroid discrepancy was 2.35 mm, 90%-mass Dice was 0.746 per channel and 0.826 per patch, and Jensen-Shannon distance was 0.330; the complete formulation improved Dice in 370 of 396 channels relative to a restricted reference. On Colin27, mean channel Dice was 0.748 and patch Dice was 0.812 across 88 retained channels. Patch-field construction required a median of only 75.81 ms. These results support CortexLume as a rapid cortical surrogate for iterative probe drafting and candidate shortlisting, with subject-specific photon modeling retained for final optical refinement.
ER -