Rat Cortical Layers Classification extracting Evoked Local Field Potential Images with Implanted Multi-Electrode Sensor

Abstract

One of the most ambitious goals of neuroscience and its neuroprosthetic applications is to interface intelligent electronic devices with the biological brain to cure neurological diseases. This emerging research field builds on our growing understanding of brain circuits and on recent technological advances in miniaturization of implantable multi-electrode-arrays (MEAs) to record brain signals at high spatiotemporal resolution. Data processing is needed to extract useful information from the recorded neural activity to better understand the function of underlying neural circuits and, in perspective, to operate neuroprosthetic devices. In this context, machine learning approaches are increasingly used in many application scenarios. This paper focuses on processing data of evoked local field potentials (LFPs) recorded from the rat barrel cortex using a miniaturized 16×16 MEA. We evaluated machine learning algorithms and trained an optimized classifier to detect at which cortical depth the neural activity is measured. We demonstrate with experimental results that machine learning can be applied successfully to noisy single-trial LFPs offering up to 99.11% of test accuracy in classifying signals acquired from different cortical layers. As such, the method is a very promising starting point toward realtime decoding of cerebral activities with low power consumption digital processors for brain-machine interfacing and neuroprosthetic applications.

Spiking neural networks provide accurate, efficient and robust models for whisker stimulus classification and allow for inter-individual generalization

“…On a chronically implanted neural prosthesis, low-powered computer architectures, designed to accurately read out, analyze and interpret brain activity, are implemented to replace the functionality of a damaged brain region. Machine learning (ML) algorithms are promising to be applied on such hardware due to their high predictive accuracy on classifying evoked neural activity as has been shown on multi-electrode array (MEA) recordings from cerebral cortex of anesthetized animals [41][42][43]. Moreover, it is of paramount importance to employ algorithms with minimal computational cost, not only to achieve real-time data processing but also for efficient hardware usage.…”

“…However, before further investigating strategies to improve the accuracy of classifying such data it might be important to consider other types of data for future work. We suggest this as the datasets that have been used so far, including the one we used, are based on precisely controlled single-whisker stimuli, certainly different from how whiskers bent and move when animals are actively palpating during unrestrained exploration [8,13,14]. Therefore, for future work, it would be highly relevant to apply and test the suggested algorithms on recordings from microchips implanted in freely moving animals.…”

“…The final subsets used for the ML-based analysis were restricted by a selection of trials for the different intensities resulting in a sample size of at least 1,000 trials, which is approximately four times higher than in comparable studies using only one recording session, e.g. [8,13,14]. Within the LOMO validation, the training set was always limited to data from three out of four mice and the size depended on the number of sessions available for the different animals.…”

“…Whisker stimulation evokes a localized response in the barrel cortex that can be captured with MEAs and analyzed with the appropriate computational approaches like machine learning (ML) classification algorithms. Based on such recordings, it has been shown that ML classification algorithms, or classifiers, can accurately identify the cortical depth of the recording electrode and the deflection intensity of the whisker stimulus [13,14]. While these authors could show that a pre-trained model can be applied on a small microchip within a reasonable amount of time, they did not provide an investigation of the capability of such hardware for training an accurate ML model online.…”

Embedded Classification of Local Field Potentials Recorded from Rat Barrel Cortex with Implanted Multi-Electrode Array

“…The whisker is deflected repeatedly by providing pulse stimuli of 5ms, while the signals are recorded from the topologically correspondent IV. STIMULATION CLASSIFICATION The decoding of the neural activities is translated into stimulation classification, following the work in [17]. The goal of this paper is to design an ML model robust to noise and able to accurately infer what kind of stimulation is received by the animal based on the LFPs recorded with the MEA sensor.…”