Related Experiment Video
Updated: May 18, 2026
12:40
Exploring Cognitive Functions in Babies, Children & Adults with Near Infrared Spectroscopy
Published on: July 28, 2009
A novel approach to probabilistic biomarker-based classification using functional near-infrared spectroscopy
Tim Hahn1, Andre F Marquand, Michael M Plichta
1Department of Cognitive Psychology II, Johann Wolfgang Goethe University Frankfurt/Main, Frankfurt am Main, Germany. TimHahn@gmx.net
Human Brain Mapping
|September 12, 2012
Summary
This study introduces a low-cost neuroimaging method using functional near-infrared spectroscopy (fNIRS) for accurate patient classification. The approach analyzes temporal patterns, overcoming limitations of expensive techniques and incorporating disease prevalence for better diagnostic predictions.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Machine Learning
Background:
- Neuroimaging analysis for patient classification faces limitations due to high costs, poor patient tolerance, and inability to incorporate prior class frequencies.
- Existing methods often rely on expensive techniques like fMRI, limiting widespread clinical application.
Purpose of the Study:
- To develop a probabilistic pattern recognition approach using cost-effective multi-channel near-infrared spectroscopy (fNIRS).
- To overcome limitations of current neuroimaging techniques by utilizing accessible fNIRS measurements and accommodating prior class frequencies.
Main Methods:
- Employed multi-channel near-infrared spectroscopy (fNIRS) for data acquisition.
- Developed a probabilistic pattern recognition algorithm integrating spatial and temporal information from fNIRS data.
- Validated the method on healthy controls for task differentiation and on schizophrenia patients and controls for classification.
Main Results:
- Successfully differentiated conditions in healthy controls using fNIRS data during a visual checkerboard task.
- Achieved high-accuracy (76%) single-subject classification of schizophrenia patients versus healthy controls using fNIRS during a working memory task.
- Demonstrated the algorithm's ability to analyze sub-second, multivariate temporal patterns of BOLD responses.
Conclusions:
- The proposed algorithm combined with fNIRS offers a low-cost, user-friendly method for high-accuracy neuroimaging-based predictions.
- This approach effectively compensates for variable class priors, enhancing its utility in diverse clinical neuroimaging applications.
- fNIRS-based pattern recognition shows significant potential for advancing diagnostic capabilities in neurological and psychiatric disorders.
