A Parametric Model for Characterizing Time-Variant Single Trials of Block-Design fNIRS Experiments.

Block-design is a popular experimental paradigm for functional near-infrared spectroscopy (fNIRS). Traditional block-design analysis techniques such as generalized linear modeling (GLM) and waveform averaging (WA) assume that the brain is a time-invariant system. This is a flawed assumption. In this paper, we propose a parametric Gaussian model to quantify the time-variant behavior found across consecutive trials of block-design fNIRS experiments. Using simulated data at different signal-to-noise ratios (SNRs), we demonstrate that our proposed technique is capable of characterizing Gaussian-like fNIRS signal features with ≥3dB SNR. When used to fit recorded data from an auditory block-design experiment, model parameter values quantitatively revealed statistically significant changes in fNIRS responses across trials, consistent with visual inspection of data from individual trials. Our results suggest that our model effectively captures trial-to-trial differences in response, which enables researchers to study time-variant brain responses using block-design fNIRS experiments.

Brain Functional Connectivity Networks do not Return to Resting-state During Control Trials in Block Design Experiments.

Many studies on morphology analysis show that if short inter-stimulus intervals separate tasks, the hemodynamic response amplitude will return to the resting-state baseline before the subsequent stimulation onset; hence, responses to successive tasks do not overlap. Accordingly, popular brain imaging analysis techniques assume changes in hemodynamic response amplitude subside after a short time (around 15 seconds). However, whether this assumption holds when studying brain functional connectivity has yet to be investigated. This paper assesses whether or not the functional connectivity network in control trials returns to the resting-state functional connectivity network. Traditionally, control trials in block-design experiments are used to evaluate response morphology to no stimulus. We analyzed data from an event-related experiment with audio and visual stimuli and resting state. Our results showed that functional connectivity networks during control trials were more similar to that of tasks than resting-state networks. In other words, contrary to task-related changes in the hemodynamic amplitude, where responses settle after a short time, the brain's functional connectivity networks do not return to their intrinsic resting-state network in such short intervals.

Impulse Data Models for the Inverse Problem of Electrocardiography.

OBJECTIVE: To develop, train and test neural networks for predicting heart surface potentials (HSPs) from body surface potentials (BSPs). The method re-frames traditional inverse problems of electrocardiography into regression problems, constraining the solution space by decomposing signals with multidimensional Gaussian impulse basis functions. METHODS: Impulse HSPs were generated with single Gaussian basis functions at discrete heart surface locations and projected to corresponding BSPs using a volume conductor torso model. Both BSP (inputs) and HSP (outputs) were mapped to regular 2D surface meshes and used to train a neural network. Predictive capabilities of the network were tested with unseen synthetic and experimental data. RESULTS: A dense full connected single hidden layer neural network was trained to map body surface impulses to heart surface Gaussian basis functions for reconstructing HSP. Synthetic pulses moving across the heart surface were predicted from the neural network with root mean squared error of 9.1±1.4%. Predicted signals were robust to noise up to 20 dB and errors due to displacement and rotation of the heart within the torso were bounded and predictable. A shift of the heart 40 mm toward the spine resulted in a 4% increase in signal feature localization error. The set of training impulse function data could be reduced, and prediction error remained bounded. Recorded HSPs from in-vitro pig hearts were reliably decomposed using space-time Gaussian basis functions. Activation times calculated from predicted HSPs for left-ventricular pacing had a mean absolute error of 10.4±11.4 ms. Other pacing scenarios were analyzed with similar success. CONCLUSION: Impulses from Gaussian basis functions are potentially an effective and robust way to train simple neural network data models for reconstructing HSPs from decomposed BSPs. SIGNIFICANCE: The HSPs predicted by the neural network can be used to generate activation maps that non-invasively identify features of cardiac electrical dysfunction and can guide subsequent treatment options.

Parametric Modeling of Electrocardiograms using Particle Swarm optimization.

The electrocardiogram (ECG) is commonly used to monitor or diagnose adverse heart conditions. While general ECG recordings are widely available, parametric ECG models have been proposed to generate ECG-like signals. Such ECG generators can create extended segments of specific beat morphology or cardiac rhythm, especially in disease states, which can be used to validate cardiac devices or evaluate ECG processing algorithms. Furthermore, ifthe parameters can be fit to a variety of ECGs, these models are valuable tools in ECG compression and modeling. In this paper we propose a framework to fit parameter values of an ECG generator such that the generated signal is similar to a reference signal. We first design a parametric ECG generator with relatively minimal assumptions of single beat waveform morphology. We then use Particle Swarm optimization to find ideal values for parameters of our ECG generator which minimize the percent root mean square difference (PRD) between the reference and generated signals. We were able to capture waveform morphologies of normal, idioventricular, and ventricular flutter rhythms with Pearson correlation coefficients above 0.9 between generated and pre-recorded signals from the MIT-BIH database.

A machine learning approach to reconstruction of heart surface potentials from body surface potentials.

Invasive cardiac catheterisation is a precursor to ablation therapy for ventricular tachycardia. Invasive cardiac diagnostics are fraught with risks. Decades of research has been conducted on the inverse problem of electrocardiography, which can be used to reconstruct Heart Surface Potentials (HSPs) from Body Surface Potentials (BSPs), for non-invasive cardiac diagnostics. State of the art solutions to the inverse problem are unsatisfactory, since the inverse problem is known to be ill-posed. In this paper we propose a novel approach to reconstructing HSPs from BSPs using a Time-Delay Artificial Neural Network (TDANN). We first design the TDANN architecture, and then develop an iterative search space algorithm to find the parameters of the TDANN, which results in the best overall HSP prediction. We use recorded BSPs and HSPs from individuals suffering from serious cardiac conditions to validate our TDANN. The results are encouraging, in that the predicted and recorded HSPs have an average correlation coefficient of 0.7 under diseased conditions.