Improving voltage-sensitive dye imaging: with a little help from computational approaches

Abstract

Voltage-sensitive dye imaging (VSDI) is a key neurophysiological recording tool because it reaches brain scales that remain inaccessible to other techniques. The development of this technique from in vitro to the behaving nonhuman primate has only been made possible thanks to the long-lasting, visionary work of Amiram Grinvald. This work has opened new scientific perspectives to the great benefit to the neuroscience community. However, this unprecedented technique remains largely under-utilized, and many future possibilities await for VSDI to reveal new functional operations. One reason why this tool has not been used extensively is the inherent complexity of the signal. For instance, the signal reflects mainly the subthreshold neuronal population response and is not linked to spiking activity in a straightforward manner. Second, VSDI gives access to intracortical recurrent dynamics that are intrinsically complex and therefore nontrivial to process. Computational approaches are thus necessary to promote our understanding and optimal use of this powerful technique. Here, we review such approaches, from computational models to dissect the mechanisms and origin of the recorded signal, to advanced signal processing methods to unravel new neuronal interactions at mesoscopic scale. Only a stronger development of interdisciplinary approaches can bridge micro- to macroscales.

Keywords: advanced signal processing; biophysical model; computational models; voltage-sensitive dye imaging.

Fig. 1

Spatio-temporal resolution and scales of neuronal recording methods. (a) Three-dimensional representation of 10 families of neuronal recording methods as a function of their spatial resolution, temporal resolution, and the spatial field-of-view that they can reach. INTRA, intracellular recordings; S/MUA, single or multiunit activity; LFP, local-field potentials; 2-PH, two-photon microscopy; MEA: multielectrode array; VSDI, voltage-sensitive dye imaging; OI-IS, optical imaging of intrinsic signals; fMRI, functional magnetic resonance imaging; E/MEG, electro- or magneto-encephalogram; IDEAL, the ideal technique. (b) In the same spatial resolution, field-of-view representation, frequency histograms of the amount of publication referenced in PubMed with the word “cortex” and one of these method. The generic search line was: (TECHNIQUE_NAME[Title/Abstract]) AND (cortex[Title/Abstract] OR cortical[Title/Abstract]). When appropriate, the technique name was written in full and abbreviated.

Fig. 2

VSDI biophysical model schematic and contributions. (a) Model representation. The six populations of neurons, depicted by one unique representative neuron (small pyramidal cells in layer 2, spiny stellate cells in layer 4, large pyramidals in layer 5, and smooth stellate cells in each layer), are recurrently connected (red arrows). The cortical column is embedded into a larger network by simulating a realistic synaptic bombardment on each population (green arrows) and by modeling lateral connections between the column and its neighbors (blue dashed arrows). Inputs from the thalamus to layer 4 neurons are represented by the large red arrow on the left. (b) Time-course of the modeled VSDI signal (red trace) in response to a thalamic input of 800 ms (black trace), compared to the experimental signal (gray trace) obtained in monkey V1. (c) Correlation analysis between the VSDI signal and the membrane potential of each compartment of the column for three periods of time (spontaneous activity, stimulation onset, or rising phase and evoked activity). (Adapted with permission from Ref. . Copyright © 2010.)

Fig. 3

Model predictions on the effect of anesthesia and experimental validation. (a) Effects of tauG modulation on the modeled VSDI signal dynamics (Plateau and DA notch amplitude). Top row: onset of the normalized VSDI signal time-courses decomposed into excitatory (burgundy traces) and inhibitory (orange traces) cells activity for three tauG values (2.5, 5, and 10 ms), revealing the DA notch formation. Middle row: time-course of the difference between excitatory and inhibitory VSDI signals. Bottom row: boxplot diagrams of the plateau amplitude (left) and the DA notch amplitude (right) of the modeled VSD response as a function of tauG. Significant differences with the condition $\mathrm{tauG}=2\mathrm{ms}$ ($P<0.01$) are denoted by a star. (b) Experimental validation of the model predictions shown in (a). Top row: onset of the experimental VSDI signal time-courses obtained in three awake (left) and three anesthetized (right) monkeys, in response to full-field drifting gratings of high contrasts. One monkey was recorded in both arousal conditions (dashed lines). Bottom row: Boxplot diagrams of the Rmax or plateau amplitude (left) and DA notch amplitude (right) values of the experimental VSDI data shown on top. (Adapted with permission from Ref. . Copyright © 2016.)

Fig. 4

Linear model decomposition. (a) A raw trial is linearly decomposed into noise components ($X_0$ baseline, $X_1$ periodic components, and $X_2$ bleaching components), the evoked response components ($X_3$), and the residual. $F$ denotes fluorescence. (b) Linear model denoising scheme. The reconstructed signal is the sum of the response components ($X_3$) and the residual, divided by the baseline illumination level to obtain a normalized reconstructed signal ($\mathrm{\Delta }F/F$). (c) Example of the linear model application on two trials in response to a-600 ms visual stimulation in the monkey visual cortex. First column: raw trials (black) and bleaching component (as estimated with the linear model; gray). Second column: other components estimated with the LM: evoked response (red) and periodic noise components (gray). Third column: residuals. Last column: estimated responses using the linear model denoising scheme (LM; red) and the standard blank subtraction (BkS; blue). Note the different scales on the ordinates axis. Adapted with permission from Ref. .

Fig. 5

Phase latency method for detecting waves in single-trial data. (a) The real, imaginary, and complex plane projections of the analytic signal (line colored by heat in time) for a damped oscillation make explicit the decomposition of a real signal into a complex phasor. (b) The complex plane projection in the previous panel is used to analyze instantaneous amplitude ($A$, complex modulus) and phase ($\phi$, complex angle) in the real signal. Gray arrows indicate the direction of phasor rotation in time. The black dot (bottom left) represents the starting point for the phase latency calculation. The blue dots represent discrete samples leading up to the phase crossing. (c) Average phase latency map for the region of interest in the primary visual cortex of awake monkey, in the stimulus (top) and black (bottom) condition. (d) Propagation speeds extracted from the slope of the relation of phase latency with distance in the unsmoothed maps, in the 50-ms stimulus condition. (Inset) Phase latency correlation with distance, stimulus (black) and blank (gray) conditions. Adapted with permission from Ref. .