In this study, we achieved a major step forward in the analysis of firing patterns of populations of motoneurons, through remarkably extensive parameter searches enabled by massively-parallel computation on supercomputers. The ability to implement these extensive parameter searches seem ideally matched to understanding the many parameters that define the inputs to neuron populations that generate these patterns. Therefore, we investigated the feasibility of using supercomputer-based models of spinal motoneurons as a basis for reverse engineering (RE) their firing patterns to identify the organization of their inputs, which we defined as the amplitudes and patterns of excitation, inhibition, and neuromodulation. This study combines two advances: 1) highly-realistic motoneuron models based on extensive in situ voltage and current clamp studies focused on neuromodulatory actions, and 2) implementation of these models using the Laboratory Computing Resource Center at Argonne National Laboratory to carry thousands (soon millions) of simulations simultaneously. The goal for computing and performing RE on over 300,000 combinations of excitatory, inhibitory, and neuromodulatory inputs was twofold: 1) to estimate the synaptic input to the motoneuron pool and 2) to generate training data for identifying the excitatory, inhibitory, and neuromodulatory inputs based on output firing patterns. As with other neural systems, any given motoneuron firing pattern could potentially be non-unique with respect to the excitatory, inhibitory, and neuromodulatory input combination (many input combinations produce similar outputs). However, our initial results show that the neuromodulatory input makes the motoneuron input-output properties so nonlinear that the effective RE solution space is restricted. The RE approach we demonstrate in this work is successful in generating estimates of the actual simulated patterns of excitation, inhibition, and neuromodulation with variances accounted for ranging from 75% to 90%. It was striking that the nonlinearities induced in firing patterns by the neuromodulation inputs did not impede RE, but instead generated distinctive features in firing patterns that aided RE. These simulations demonstrate the potential of this form of RE analysis. It is likely that the ever-increasing power of supercomputers will allow increasingly accurate RE of neuron inputs from their firing patterns from many neural systems.

The results show that a NN could make lesion depth estimations in real-time using less in situ devices than current techniques. With the NN-based technique, physicians could deliver quicker and more precise ablation therapy.

Background: Radiofrequency ablation is a minimally-invasive treatment method that aims to destroy undesired tissue by exposing it to alternating current in the 100 kHz-800 kHz frequency range and heating it until it is destroyed via coagulative necrosis. Ablation treatment is gaining momentum especially in cancer research, where the undesired tissue is a malignant tumor. While ablating the tumor with an electrode or catheter is an easy task, real-time monitoring the ablation process is a must in order to maintain the reliability of the treatment. Common methods for this monitoring task have proven to be accurate, however, they are all time-consuming or require expensive equipment, which makes the clinical ablation process more cumbersome and expensive due to the time-dependent nature of the clinical procedure. Methods: A machine learning (ML) approach is presented that aims to reduce the monitoring time while keeping the accuracy of the conventional methods. Two different hardware setups are used to perform the ablation and collect impedance data at the same time and different ML algorithms are tested to predict the ablation depth in 3 dimensions, based on the collected data. Results: Both the random forest and adaptive boosting (adaboost) models had over 98% R 2 on the data collected with the embedded system-based hardware instrumentation setup, outperforming Neural Network-based models. Conclusions: It is shown that an optimal pair of hardware setup and ML algorithm (Adaboost) is able to control the ablation by estimating the lesion depth within a test average of 0.3mm while keeping the estimation time within 10ms on a Â86-64 workstation.

One of the most common types of models that helps us to understand neuron behavior is based on the Hodgkin–Huxley ion channel formulation (HH model). A major challenge with inferring parameters in HH models is non-uniqueness: many different sets of ion channel parameter values produce similar outputs for the same input stimulus. Such phenomena result in an objective function that exhibits multiple modes (i.e., multiple local minima). This non-uniqueness of local optimality poses challenges for parameter estimation with many algorithmic optimization techniques. HH models additionally have severe non-linearities resulting in further challenges for inferring parameters in an algorithmic fashion. To address these challenges with a tractable method in high-dimensional parameter spaces, we propose using a particular Markov chain Monte Carlo (MCMC) algorithm, which has the advantage of inferring parameters in a Bayesian framework. The Bayesian approach is designed to be suitable for multimodal solutions to inverse problems. We introduce and demonstrate the method using a three-channel HH model. We then focus on the inference of nine parameters in an eight-channel HH model, which we analyze in detail. We explore how the MCMC algorithm can uncover complex relationships between inferred parameters using five injected current levels. The MCMC method provides as a result a nine-dimensional posterior distribution, which we analyze visually with solution maps or landscapes of the possible parameter sets. The visualized solution maps show new complex structures of the multimodal posteriors, and they allow for selection of locally and globally optimal value sets, and they visually expose parameter sensitivities and regions of higher model robustness. We envision these solution maps as enabling experimentalists to improve the design of future experiments, increase scientific productivity and improve on model structure and ideation when the MCMC algorithm is applied to experimental data.

No abstract

No abstract

No abstract

Experimental data-based parameter search for Hodgkin–Huxley-style (HH) neuron models is a major challenge for neuroscientists and neuroengineers. Current search strategies are often computationally expensive, are slow to converge, have difficulty handling nonlinearities or multimodalities in the objective function, or require good initial parameter guesses. Most important, many existing approaches lack quantification of uncertainties in parameter estimates even though such uncertainties are of immense biological significance. We propose a novel method for parameter inference and uncertainty quantification in a Bayesian framework using the Markov chain Monte Carlo (MCMC) approach. This approach incorporates prior knowledge about model parameters (as probability distributions) and aims to map the prior to a posterior distribution of parameters informed by both the model and the data. Furthermore, using the adaptive parallel tempering strategy for MCMC, we tackle the highly nonlinear, noisy, and multimodal loss function, which depends on the HH neuron model. We tested the robustness of our approach using the voltage trace data generated from a 9-parameter HH model using five levels of injected currents (0.0, 0.1, 0.2, 0.3, and 0.4 nA). Each test consisted of running the ground truth with its respective currents to estimate the model parameters. To simulate the condition for fitting a frequency-current (F-I) curve, we also introduced an aggregate objective that runs MCMC against all five levels simultaneously. We found that MCMC was able to produce many solutions with acceptable loss values (e.g., for 0.0 nA, 889 solutions were within 0.5% of the best solution and 1,595 solutions within 1% of the best solution). Thus, an adaptive parallel tempering MCMC search provides a “landscape” of the possible parameter sets with acceptable loss values in a tractable manner. Our approach is able to obtain an intelligently sampled global view of the solution distributions within a search range in a single computation. Additionally, the advantage of uncertainty quantification allows for exploration of further solution spaces, which can serve to better inform future experiments.