Motometrics: A Toolbox for Annotation and Efficient Analysis of Motor Evoked Potentials

Abstract

Stimulating the nervous system and measuring muscle response offers a unique opportunity to interrogate motor system function. Often, this is performed by stimulating motor cortex and recording muscle activity with electromyography; the evoked response is called the motor evoked potential (MEP). To understand system dynamics, MEPs are typically recorded through a range of motor cortex stimulation intensities. The MEPs increase with increasing stimulation intensities, and these typically produce a sigmoidal response curve. Analysis of MEPs is often complex and analysis of response curves is time-consuming. We created an MEP analysis software, called Motometrics, to facilitate analysis of MEPs and response curves. The goal is to combine robust signal processing algorithms with a simple user interface. Motometrics first enables the user to annotate data files acquired from the recording system so that the responses can be extracted and labeled with the correct subject and experimental condition. The software enables quick visual representations of entire datasets, to ensure uniform quality of the signal. It then enables the user to choose a variety of response curve analyses and to perform near real time quantification of the MEPs for quick feedback during experimental procedures. This is a modular open source tool that is compatible with several popular electrophysiological systems. Initial use indicates that Motometrics enables rapid, robust, and intuitive analysis of MEP response curves by neuroscientists without programming or signal processing expertise.

Keywords: EMG; MEP; analysis; evoked; motor; recruitment; software.

Figure 1

Generation of recruitment curves during an experimental session: Stimulation is applied to the motor cortex with varying intensities. Corresponding MEPs are recorded and saved as recruitment curve session files by the electrophysiology system.

Figure 2

Process overview: Motometrics uses modules to organize experimental data, quantify MEPs, fit recruitment curves, and compare recruitment curves across conditions. (A) The user organizes and annotates MEP data files relevant for the generation of recruitment curves. (B) The user inspects the quality of data and removes data with artifacts. The user may also select to filter the data to remove noise and correct baseline drift. (C) The user then chooses a metric, such as area under the curve or peak amplitude, to quantify the organized MEP data. (D) Recruitment curves are generated using a curve fitting algorithm. (E) The user next selects a recruitment curve metric (e.g., slope) to quantify and compare across curves.

Figure 3

Organization of data. (A) A single experimental recording consists of MEPs obtained using a defined range of stimulation intensities. At every stimulus intensity multiple trials are obtained. The data is saved as a multidimensional matrix in a MATLAB data file. (B) At the next level of data organization, we specify multiple experimental recordings in a record file for a single subject. Here, we utilize experiment labels to denote different experimental conditions for that subject. The software can use a record file to generate and analyze recruitment curves for a single subject. (C) Finally we use a master file to group multiple subjects or data from the same subject across time with similar experimental conditions for comparison of recruitment curves. The software can use a master file to enable detailed comparisons across multiple subjects.

Figure 4

Recruitment curve metrics. A set of recruitment curves from a single record file may be evaluated using 3 different metrics. (A) The MEP metric enables recruitment curves to compare their quantified MEP responses for a fixed stimulation. (B) The Stimulation metric compares stimulation intensities required to elicit a certain value of the quantified MEP. (C) The Slope metric estimates the tangent of the angle made by the linear portion of the recruitment curve with the horizontal line.

Figure 5

Motometrics wizard. The wizard guides the novice or unfamiliar user in selecting relevant parameter choices for configuring Motometrics. Each parameter is explained and acceptable ranges of values are suggested.

Figure 6

Software architecture: Motometrics uses a 3 layered software architecture model to achieve a modular design. The interface layer contains MATLAB function modules consisting of GUI windows to accept user input and provides graphical output to the user. The Data annotation modules are a part of this layer. The processing layer consists of the back end modules that extract and process MEP data to generate quantified MEPs and recruitment curves. The modules in yellow boxes correspond to the conceptual modules listed in Figure 2. The data layer represents the various data files and data structures that Motometrics reads and saves to for further analysis.

Figure 7

Motometrics record manager. The Record Manager enables one to annotate and group session data. (A) The table on the left, enables the user to input annotation information such as a unique session ID that corresponds to a specific experimental condition. This is generated automatically from the parameters on the right, but these can be changed within the table. (B) The controls on the right assist in the annotation process by identifying the session ID number and stimulation strength. Keyboard shortcuts can be used for faster annotation. (C) The list on the bottom right enables each session ID/experimental condition to be linked to the recruitment curve session file to create the record file. Previously created record files may also be loaded later for editing.

Figure 8

Data selection and visualization tools. (A) Motometrics provides a list of record files that can be processed with the data exploration tool by double clicking on any of them. (B) The data explorer tool enables the user to visualize MEP data or screen for artifacts. (C) The Visualization tool enables the user to examine MEP data that are extracted from the MEP session files specified in a record. (D) The artifact screening tool enables the user to obtain a graphical overview of all MEP data of a recruitment curve session in a single figure. The heat map representation makes it easy for the user to eyeball the data in order to spot inconsistencies or artifacts.

Figure 9

Quantifying MEPs and visualizing recruitment curves. (A) This GUI in the window of Motometrics window provides choices for different MEP metrics. (B) After the choice of MEP quantification metric is made, and the Process Data Files button is clicked, recruitment curves are generated.

Figure 10

Analyzed recruitment curves. Recruitment curves are normalized to a baseline reference and displayed. (A) The MEP50 metric compares MEP magnitudes of recruitment curves given the stimulation intensity that produces 50% of maximum baseline MEP. (B) The Stim50 metric compares the amount of stimulation required for each recruitment curve to produce a MEP magnitude equal to the Baseline recruitment curve when 50% of maximum stimulation is used. (C) The Slope metric compares the rate of MEP increase with increasing stimulation intensity.