Ultra-high density imaging arrays for diffuse optical tomography of human brain improve resolution, signal-to-noise, and information decoding

Abstract

Functional magnetic resonance imaging (fMRI) has dramatically advanced non-invasive human brain mapping and decoding. Functional near-infrared spectroscopy (fNIRS) and high-density diffuse optical tomography (HD-DOT) non-invasively measure blood oxygen fluctuations related to brain activity, like fMRI, at the brain surface, using more-lightweight equipment that circumvents ergonomic and logistical limitations of fMRI. HD-DOT grids have smaller inter-optode spacing (~13 mm) than sparse fNIRS (~30 mm) and therefore provide higher image quality, with spatial resolution ~1/2 that of fMRI. Herein, simulations indicated reducing inter-optode spacing to 6.5 mm would further improve image quality and noise-resolution tradeoff, with diminishing returns below 6.5 mm. We then constructed an ultra-high-density DOT system (6.5-mm spacing) with 140 dB dynamic range that imaged stimulus-evoked activations with 30-50% higher spatial resolution and repeatable multi-focal activity with excellent agreement with participant-matched fMRI. Further, this system decoded visual stimulus position with 19-35% lower error than previous HD-DOT, throughout occipital cortex.

Identifying Naturalistic Movies from Human Brain Activity with High-Density Diffuse Optical Tomography

“…This finding was consistent with previous work indicating superior image quality and higher template-based decoding performance for UHD-DOT than its HD-subset. 22 We also performed a simpler, template-based decoding task, in which we designated trials in the first and second half of each session as training and test trials, respectively, block-averaged the training trials to form “templates”, and used a maximum-correlation classifier to guess which of the 4 clips was viewed in each test trial ( Supplementary Figs. S1, S6B ).…”

“…That sensitivity matrix was computed from finite-element simulations of light propagation through the head of a single subject, a segmented 5-layer model of scalp, skull, cerebrospinal fluid, gray matter, and white matter tissue obtained from anatomical T1-weighted and T2-weighted MRI head images acquired in a previous study. 22 Next, the absorption coefficient fluctuation images at each light wavelength were spectroscopically converted into concentration fluctuations in oxyhemoglobin and deoxyhemoglobin. Finally, to further isolate brain-specific signals, voxels in scalp and skull were excluded from encoding model fitting and clip identity decoding analyses.…”

“…To image hemodynamic signals from occipital brain areas sensitive to visual stimuli, we placed an ultra-high-density DOT system with 126 sources and 126 detectors covering a 12cm x 5.5cm area on the back of the head, 22 and we employed well-established signal processing and image reconstruction pipelines for DOT. 13,15,33,34 Briefly, we converted measured light level time traces in each measurement channel (i.e., each source-detector pair and light wavelength) into log-ratio time traces representing the fractional change in light level about its mean in that channel.…”

“…To perform a more-basic, template-based decoding task similar to previous HD-DOT decoding 18,22 as a validation and comparison, we treated the data from the first and second halves of each imaging session as training and testing sets, then block-averaged the responses to each clip to serve as “templates,” then had a decoder guess that the clip being viewed was the one whose template response was most highly correlated with the actual response in each test trial. We repeated many of the manipulations described elsewhere in this paper to evaluate template-based decoding performance for between-session training and testing, decreasing time window duration, increasing number of clips to decode, and reducing optode grid density ( Fig.…”

“…1C-E). 22 To evaluate the repeatability and similarity of the brain's responses to each clip, we computed the correlation between the average brain response to each clip from the first and second half of each imaging session (Supplementary Figs. S1, S2).…”