|
|
Data processing in real-time fMRI neurofeedback
Bečička, Martin ; Slavíček, Tomáš (oponent) ; Lamoš, Martin (vedoucí práce)
The presented thesis deals with real-time digital filtering of fMRI neurofeedback data. It analyzes currently used solution at CEITEC MU chiefly in respect to finding ways to shorten the delay at the beginning of each neurofeedback block which is introduced by digital filtering. Current solution uses extended Kalman filter mainly for its real-time and smoothing properties. Analysis of 150 individual neurofeedback blocks yielded true learning period of Kalman filter which has been found to be significantly shorter than is set in the current solution. Different options to further reduce the transient period have been explored and short moving average filter has been chosen as an optimal trade-off between transient period, filter delay and its smoothing properties.
|
|
| |
|
|
Evaluation of selected thermophysiological model for the isolated, confined mission in Antarctica
Bečička, Martin ; Dobrovolná,, Julie (oponent) ; Paštěka, Richard (vedoucí práce)
Over the past few decades, numerous models have been created to estimate human thermal responses. These models rely on energy balance equations that consider the heat exchange between the human body and its environment. These models have substantially progressed from one-node models into complex structures organized in multiple layers and connected by a circulatory blood flow. Although there is literature available comparing the possible utility of the thermophysiological models in different scenarios, little is known about the utility of these mod els in specific conditions. This work investigates and compares selected thermophysiological models, focusing on selecting a suitable model for the use case scenario of Antarctica. The data was collected during a longitudinal study located at Johann Gregor Mendel Czech Antarctic station. The local skin temperature of participants (n = 16) was measured 6 times over the period of the study. The first of each measurement set consisted of orthostatic tests in addition to psychomotor vigilance task and Iowa gambling test present which were present in each of the subsequent measurements. After examining specific experimental conditions, four thermophysiological models were com pared, and the JOS-3 model was determined to be the most appropriate for the thermal data obtained during the experiment. The results revealed that the selected JOS-3 model can accurately predict mean skin temper atures for both experiment designs (RMSD 0.72 and 0.75 C for experiment design 1 and 2 respectively). Furthermore, the JOS-3 model was able to reliably predict head (RMSD 0.82 and 1.13 C), neck (RMSD 0.61 and 1 C), and chest (RMSD 1.01 and 1.22 C) skin temperatures. However, deviations observed for back, hand, and foot segments were significant (RMSD 2.08 - 2.25 and 2.55 - 2.68 C). Possible explanations include uncertainty in clothing insulation of the worn clothing, uncertainty in measurement circumstances (contact with surfaces - floor, table), and the placement of the thermal sensors. The highest deviation was observed in the pelvis (RMSD 2.25 and 3.72 C) and is possibly a result of the JOS-3 model overestimation of heat production in the segment. The study has underscored the significance of obtaining compre hensive data on clothing type and sensor placement for enhancing the accuracy of simulation outcomes.
|
|
|
Data processing in real-time fMRI neurofeedback
Bečička, Martin ; Slavíček, Tomáš (oponent) ; Lamoš, Martin (vedoucí práce)
The presented thesis deals with real-time digital filtering of fMRI neurofeedback data. It analyzes currently used solution at CEITEC MU chiefly in respect to finding ways to shorten the delay at the beginning of each neurofeedback block which is introduced by digital filtering. Current solution uses extended Kalman filter mainly for its real-time and smoothing properties. Analysis of 150 individual neurofeedback blocks yielded true learning period of Kalman filter which has been found to be significantly shorter than is set in the current solution. Different options to further reduce the transient period have been explored and short moving average filter has been chosen as an optimal trade-off between transient period, filter delay and its smoothing properties.
|
|
| |
host ::
RSS.