An intuitive and generalisable approach to spatial-temporal feature extraction for high-density (HD) functional Near-Infrared Spectroscopy (fNIRS) brain-computer interface (BCI) is proposed, demonstrated here using Frequency-Domain (FD) fNIRS for motor-task classification. Enabled by the HD probe design, layered topographical maps of Oxy/deOxy Haemoglobin changes are used to train a 3D convolutional neural network (CNN), enabling simultaneous extraction of spatial and temporal features. The proposed spatial-temporal CNN is shown to effectively exploit the spatial relationships in HD fNIRS measurements to improve the classification of the functional haemodynamic response, achieving an average F1 score of 0.69 across seven subjects in a mixed subjects training scheme, and improving subject-independent classification as compared to a standard temporal CNN.INDEX TERMS fNIRS, brain-computer interface, neural network, machine learning, CNN IMPACT STATEMENT The unique and often overlooked interdependence between fNIRS probe design and neural network architecture is demonstrated. A novel and effective BCI model is presented, which can be generalised to multi-wavelength, broadband, and hybrid fNIRS data.

No abstract

Frequency domain (FD) diffuse optical spectroscopy (DOS) can be used to recover absolute optical properties of biological tissue, providing valuable clinical feedback, including in diagnosis and monitoring of breast tumours. In this study, tomographic (3D) and topographic (2D) techniques for spatially-varying optical parameter recovery are presented, based on a multi-distance, handheld DOS probe. Processing pipelines and reconstruction quality are discussed and quantitatively compared, demonstrating the trade-offs between depth sensitivity, optical contrast, and computational speed. Together, the two techniques provide both depth sensitive real-time feedback, and high-resolution 3D reconstruction from a single set of measurements, enabling faster and more accurate clinical feedback.

Functional Near-Infrared Spectroscopy (fNIRS) is a non-invasive, non-ionizing imaging tool that can map brain hemodynamics. While not the most common fNIRS approach, frequency-domain NIRS (FD-NIRS) has shown an ability to estimate the absolute optical properties of tissues and, consequently, accurately estimate tissue chromophores concentrations. FD-NIRS can probe different depths in the tissue using multiple source-detector separations (multi-distance) or multiple modulation frequencies (multi-frequency). In this work, through experimental and simulation results, we demonstrate that using multi-distance and multi-frequency FD-NIRS yields similar results when estimating the optical properties of homogeneous and multi-layered tissues with less than ±10% error in estimations. We also examined some parameters that can affect the accuracy of the estimated optical properties, such as using different modulation frequencies in a multi-distance configuration and different source-detector separations for multi-frequency configuration.

No abstract