Using deep neural networks to detect complex spikes of cerebellar Purkinje Cells

Abstract

<jats:title>Abstract</jats:title><jats:p>One of the most powerful excitatory synapses in the entire brain is formed by cerebellar climbing fibers, originating from neurons in the inferior olive, that wrap around the proximal dendrites of cerebellar Purkinje cells. The activation of a single olivary neuron is capable of generating a large electrical event, called “complex spike”, at the level of the postsynaptic Purkinje cell, comprising of a fast initial spike of large amplitude followed by a slow polyphasic tail of small amplitude spikelets. Several ideas discussing the role of the cerebellum in motor control are centered on these complex spike events. However, these events are extremely rare, only occurring 1-2 times per second. As a result, drawing conclusions about their functional role has been very challenging, as even few errors in their detection may change the result. Since standard spike sorting approaches cannot fully handle the polyphasic shape of complex spike waveforms, the only safe way to avoid omissions and false detections has been to rely on visual inspection of long traces of Purkinje cell recordings by experts. Here we present a supervised deep learning algorithm for rapidly and reliably detecting complex spikes as an alternative to tedious visual inspection. Our algorithm, utilizing both action potential and local field potential signals, not only detects complex spike events much faster than human experts, but it also excavates key features of complex spike morphology with a performance comparable to that of such experts.</jats:p><jats:sec><jats:title>Significance statement</jats:title><jats:p>Climbing fiber driven “complex spikes”, fired at perplexingly low rates, are known to play a crucial role in cerebellum-based motor control. Careful interpretations of these spikes require researchers to manually detect them, since conventional online or offline spike sorting algorithms (optimized for analyzing the much more frequent “simple spikes”) cannot be fully trusted. Here, we present a deep learning approach for identifying complex spikes, which is trained on local field and action potential recordings from cerebellar Purkinje cells. Our algorithm successfully identifies complex spikes, along with additional relevant neurophysiological features, with an accuracy level matching that of human experts, yet with very little time expenditure.</jats:p></jats:sec>

P-sort: an open-source software for cerebellar neurophysiology

“…P-sort utilizes a novel dimensionality reduction algorithm called UMAP (McInnes et al, 2018). UMAP is a nonlinear technique that, in our experience, is particularly powerful for clustering waveforms and identifying complex spikes, as also shown by the work of Markanday et al (2020). Indeed, in the case of the data in Fig.…”

“…Identification of complex spikes suffers from additional problems. There are variable number of spikelets in the complex spike waveform (Burroughs et al, 2017; Davie et al, 2008; Ito & Simpson, 1971; Monsivais et al, 2005; Najafi & Medina, 2013; Yang & Lisberger, 2014), and thus template matching may have difficulty labeling all complex spikes (Markanday et al, 2020). Examples of the variable spikelets are shown in Fig.…”

“…Finally, the complex spike waveform can be distorted because of the proximity of a simple spike, as shown in Fig. 2B (Markanday et al, 2020; Servais et al, 2004). This is because simple and complex spikes are driven by different inputs to the P-cell, one is from a climbing fiber that generates a dendritic complex spike, while the other is from granule cells that generate somatic simple spikes.…”

“…As a result, a typical complex spike can produce a “broad spike” in the low-pass filtered representation of the data (local field potential, LFP). Indeed, two recent developments in complex spike detection are novel algorithms that depend partly on the LFP waveform (Markanday et al, 2020; Zur & Joshua, 2019). Once the simple and complex spikes are labeled, the final step is to determine whether the simple spikes have been suppressed after a complex spike.…”

Non-linear Dimensionality Reduction on Extracellular Waveforms Reveals Cell Type Diversity in Premotor Cortex

“…These issues led us to investigate techniques that don't require feature specification but are designed to find patterns in complex datasets through non-linear dimensionality reduction. Such methods have seen usage in diverse neuroscientific contexts such as single-cell transcriptomics 37,38 , in analyzing models of biological neural networks 39,40 , the identification of behavior [41][42][43] , and in electrophysiology [44][45][46][47] .…”