Validating silicon polytrodes with paired juxtacellular recordings: method and dataset

Abstract

Cross-validating new methods for recording neural activity is necessary to accurately interpret and compare the signals they measure. Here we describe a procedure for precisely aligning two probes for in vivo "paired-recordings" such that the spiking activity of a single neuron is monitored with both a dense extracellular silicon polytrode and a juxtacellular micropipette. Our new method allows for efficient, reliable, and automated guidance of both probes to the same neural structure with micrometer resolution. We also describe a new dataset of paired-recordings, which is available online. We propose that our novel targeting system, and ever expanding cross-validation dataset, will be vital to the development of new algorithms for automatically detecting/sorting single-units, characterizing new electrode materials/designs, and resolving nagging questions regarding the origin and nature of extracellular neural signals.

Keywords: extracellular action potential; ground truth; juxtacellular recording; polytrodes; spike sorting.

Fig. 1.

In vivo paired-recording setup: design and method. A: schematic of the dual-probe recording station. The “PatchStar” (PS) micromanipulator drives the juxtacellular pipette and the “In-Vivo Manipulator” (IVM) manipulator drives the extracellular polytrode. The setup includes a long working distance microscope assembled from optomechanical components mounted on a 3-axis motorized stage. The alignment image provides a high-resolution view from above the stereotactic frame, top left; however, a side-view can also be obtained for calibration purposes, top right (scale bar = 100 μm). B: schematic of a coronal view [reproduced from Paxinos G, Watson C. The Rat Brain In Stereotaxic Coordinates (6th ed.), Elsevier, 2007, with permission] of the craniotomy and durotomies with both probes positioned at the calibration point. The distance between durotomies, such that the probe tips meet at deep layers in cortex, was ∼2 mm. The black arrows represent the motion path for both electrodes entering the brain (scale bar = 1 mm). C: diagram of simultaneous extracellular and juxtacellular paired-recording of the same neuron at a distance of 90 μm between the micropipette tip and the closest electrode on the extracellular polytrode (scale bar = 100 μm).

Fig. 2.

Paired extracellular and juxtacellular recordings from the same neuron. A: recording pipette with a long thin taper used for juxtacellular recordings with typical tip diameter of 1–4 μm and resistance of 3–7 MΩ. B: representative juxtacellular recording from a cell ∼1,256 μm in depth, 51 μm from the extracellular probe (2014_10_17_Pair1.0), with a firing rate of ∼ 1 Hz. C: juxtacellular action potentials are overlaid, time-locked to the maximum positive peak, with the average spike waveform superimposed (n = 442 spikes). D: extracellular dense polytrode array with a span of 275 μm along the shank axis; the electrode channel number is represented at each site. E: representative extracellular recording that corresponds to the same time window as the above juxtacellular recording. Traces are ordered from upper to lower electrodes and channel numbers are indicated. F: extracellular waveforms, aligned on the juxtacellular spike peak, for a single channel (channel 18) and the juxtacellular triggered average (JTA) obtained by including an increasing number of juxtacellular events (n as indicated). G: spatial distribution of the amplitude for each channel's extracellular JTA waveform. The peak-to-peak amplitude within a time window (±1 ms) surrounding the juxtacellular event was measured and the indicated color code was used to display and interpolate these amplitudes throughout the probe shaft. H: waveform averages for all the extracellular electrodes are spatially arranged. The channel with the highest peak-to-peak JTA amplitude (channel 18) is marked with a black asterisk and the closest channel (channel 9) is marked with a red asterisk. I: extracellular JTA time courses for each channel are overlaid and colored according to the scheme in H.

Fig. 3.

Distance dependence of extracellular signal amplitude. The maximum peak-to-peak amplitude of the JTAs (±1 ms of the alignment time) across all extracellular channels for each paired-recording vs. the distance between the closest extracellular electrode and the juxtacellular pipette tip. Horizontal error bars report uncertainty in position estimate (±10.5 μm). The gray shaded region indicates a 5-μV threshold for excluding possible cross-talk electrical artifacts between the extra- and juxtacellular recording electronics.

Fig. 4.

Extracellular detection of the juxtacellular neuron's action potentials. Representative juxtacellular recording (A) and wide-band (B; 0.1–7,500 kHz) signal recorded simultaneously with a 32-channel silicon polytrode. C: on the highpass filtered extracellular data is visible the occurrence of temporally overlapping spikes on separated electrodes. The highlighted traces are expanded in the right panel and included black arrows to indicate all spikes identified by SpikeDetekt using standard thresholds and green arrows to indicate the time of juxtacellular spikes. D: peri-event time histograms of the extracellular spike events found by SpikeDetekt, relative to the juxtacellular spike times in 1-ms bins centered at 0 ms, are shown for each electrode channel at their relative position on the extracellular probe. The channel with the largest peak in the bin at 0 (±0.5 ms from the juxtacellular event) is indicated by asterisk and expanded at bottom. E: same presentation as in D but for a neuron with a smaller extracellular action potential.

Fig. 5.

Spatiotemporal structure of extracellular signatures. A: geometry and dimensions of the 32-channel electrode array. B: JTA waveforms (2014_11_25_Pair1.1) for all the extracellular electrodes are spatially arranged according to the probe geometry. C: expanded comparison of the JTA waveforms for the indicated electrodes with a line denoting the peak time of the juxtacellular spike. D–F: similar presentation as (A–C) for one 128-channel polytrode pair example (2015_09_04_Pair5.0).

Fig. 6.

Dataset for validating spike detection and sorting algorithms for dense polytrodes. A: spatial distribution of the peak-to-peak amplitude within a time window (±1 ms) surrounding the juxtacellular event and the indicated color code was used to display and interpolate these amplitudes throughout the 32-channel probe shaft. In addition, the extracellular JTA waveforms for all the extracellular electrodes are spatially arranged. B: same presentation as in A for paired-recordings with the 128-channel probe.