Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Implementation of a Doublet-producing Leaky Integrate-and-Fire Motor Neuron Membrane Model : Comparing Simulated Human Motor Neuron Behavior with Data from Surface Electromyography

Abeln, Lilly LU and Bowers, Joseph Rylan LU (2025) BMEM01 20251
Department of Biomedical Engineering
Abstract
The human central nervous system is highly complex, and doublet discharge (two signal spikes in unusually short succession) behavior observed in motor neuron output remains only partially understood. Doublets play a key role in rapid force production during muscle activation. While computational models that capture this behavior exist, their complexity is too computationally demanding for use outside of academic research and makes any modifications difficult. This thesis addresses the need for a simpler linear model by developing a leaky integrate-and-fire (LIF) model producing doublets while maintaining physiologically plausible firing frequencies across a range of input conditions. The primary goal of this project was to construct and... (More)
The human central nervous system is highly complex, and doublet discharge (two signal spikes in unusually short succession) behavior observed in motor neuron output remains only partially understood. Doublets play a key role in rapid force production during muscle activation. While computational models that capture this behavior exist, their complexity is too computationally demanding for use outside of academic research and makes any modifications difficult. This thesis addresses the need for a simpler linear model by developing a leaky integrate-and-fire (LIF) model producing doublets while maintaining physiologically plausible firing frequencies across a range of input conditions. The primary goal of this project was to construct and evaluate how well such a model can replicate experimentally observed doublet behavior.

Three models were implemented and evaluated against literature on motor neuron characteristics, and experimental surface EMG data from two subjects. Model 1 explored a threshold-based firing mechanism with reactive inhibition, producing doublets at neuron activation but not repetitive occurrences. Model 2 mimicked physiological modulation of neuron excitability (delayed depolarization), but overproduced doublets. Model 3 combined both mechanisms, balancing excitability with dynamic inhibition to reproduce a broader and physiologically realistic range of doublet patterns.

Although the models did not perfectly replicate experimental data, biological variability between individuals makes a universal solution unfeasible. The models produced a satisfactory range of firing patterns under varied input conditions while maintaining a simple underlying structure. The thesis also highlights limitations regarding uniform parameters and simplified connectivity in neuron pool modeling. Future work could introduce neuron- or cluster-specific dynamics to better simulate diverse neuron profiles. This work demonstrates the viability of a simple LIF model for reproducing physiologically reasonable behavior, including doublets, under various input conditions, while remaining flexible enough to incorporate additional features. (Less)
Popular Abstract
From Noise to Signal: Rediscovering the Doublet

Neurons relay signal patterns for motor control. Some signals could help diagnose ALS earlier, but are currently intermingled with and lost in the noise.

Neurons are the information highway of the body, carrying instructions for motor control from brain to muscle. Understanding these signals is important for research on neuromuscular diseases.

Poking needles into people to see what their neurons are saying is as uncomfortable as it sounds. Surface electromyography (sEMG) measurements can detect neuron signals using a patch with many electrodes in contact with the skin. Since these
electrodes are exposed to both external signals and noise from other neighbouring tissues/muscles,... (More)
From Noise to Signal: Rediscovering the Doublet

Neurons relay signal patterns for motor control. Some signals could help diagnose ALS earlier, but are currently intermingled with and lost in the noise.

Neurons are the information highway of the body, carrying instructions for motor control from brain to muscle. Understanding these signals is important for research on neuromuscular diseases.

Poking needles into people to see what their neurons are saying is as uncomfortable as it sounds. Surface electromyography (sEMG) measurements can detect neuron signals using a patch with many electrodes in contact with the skin. Since these
electrodes are exposed to both external signals and noise from other neighbouring tissues/muscles, advanced signal processing is required to disentangle the measurement. Current noise-filtering strategies involve decomposition into individual spike-trains based on a narrow frequency range, however this results in some collateral damage.

In certain situations, such as when your body wants to move quickly, your neurons may send double-signal spikes to your muscles. These so-called doublets produce more force than two regular consecutive signal spikes would, but due to the relatively high instantaneous frequency of these doublets, they are caught in the high-frequency noise filtering as part of EMG signal processing. Important information on force modulation is then lost in the process.

Excessive production of doublets is also associated with ALS, and is an early sign of disease progression, possibly even before symptoms have begun to manifest. Here lies an unrealised potential for earlier diagnosis, which a better understanding of doublets and their underlying mechanisms could enable.

The difficulty of detecting and studying doublets makes computational modelling in combination with sEMG an attractive option, reducing the need for invasive measurement techniques. Current computational models either cannot capture doublets, or rely on difficult to obtain human data on underlying processes for signal production. This work aimed to build a simpler neuron model that could replicate doublet behaviour based on a literature and model review, and sEMG data from two healthy subjects.

The general behaviour of doublets was successfully captured through conditional inhibition and excitability of the neuron, with each model iteration moving closer to experimentally observed behaviour. The introduction of parameter variability and grouping of neurons is the next stage for development of this model, which serves as a framework on which further features can be built.

Doublets are an important part of motor control and the ability to model these signals could help fill gaps in motor neuron research. To mimic human motion, we first need to understand what is controlling it, something this model has the potential to contribute to with further development. (Less)
Please use this url to cite or link to this publication:
author
Abeln, Lilly LU and Bowers, Joseph Rylan LU
supervisor
organization
course
BMEM01 20251
year
type
H2 - Master's Degree (Two Years)
subject
keywords
motoneuron, motor neuron, surface electromyography (sEMG), doublet, neuron signals
language
English
additional info
2025-11
id
9195429
date added to LUP
2025-06-23 12:28:07
date last changed
2025-06-23 12:28:07
@misc{9195429,
  abstract     = {{The human central nervous system is highly complex, and doublet discharge (two signal spikes in unusually short succession) behavior observed in motor neuron output remains only partially understood. Doublets play a key role in rapid force production during muscle activation. While computational models that capture this behavior exist, their complexity is too computationally demanding for use outside of academic research and makes any modifications difficult. This thesis addresses the need for a simpler linear model by developing a leaky integrate-and-fire (LIF) model producing doublets while maintaining physiologically plausible firing frequencies across a range of input conditions. The primary goal of this project was to construct and evaluate how well such a model can replicate experimentally observed doublet behavior.

Three models were implemented and evaluated against literature on motor neuron characteristics, and experimental surface EMG data from two subjects. Model 1 explored a threshold-based firing mechanism with reactive inhibition, producing doublets at neuron activation but not repetitive occurrences. Model 2 mimicked physiological modulation of neuron excitability (delayed depolarization), but overproduced doublets. Model 3 combined both mechanisms, balancing excitability with dynamic inhibition to reproduce a broader and physiologically realistic range of doublet patterns.

Although the models did not perfectly replicate experimental data, biological variability between individuals makes a universal solution unfeasible. The models produced a satisfactory range of firing patterns under varied input conditions while maintaining a simple underlying structure. The thesis also highlights limitations regarding uniform parameters and simplified connectivity in neuron pool modeling. Future work could introduce neuron- or cluster-specific dynamics to better simulate diverse neuron profiles. This work demonstrates the viability of a simple LIF model for reproducing physiologically reasonable behavior, including doublets, under various input conditions, while remaining flexible enough to incorporate additional features.}},
  author       = {{Abeln, Lilly and Bowers, Joseph Rylan}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Implementation of a Doublet-producing Leaky Integrate-and-Fire Motor Neuron Membrane Model : Comparing Simulated Human Motor Neuron Behavior with Data from Surface Electromyography}},
  year         = {{2025}},
}