The hemodynamic model describes the blood flow mechanisms as well as the coupling between the metabolic and neuronal activities in a voxel on the human brain. The solution and deconvolution for identifying dynamical variables, biophysical parameters, and hidden states of this hemodynamic model is a researchable work. In order to solve the hemodynamic model parameters, hidden states, and estimate blood oxygen level dependent (BOLD) signal from fMRI images, there are three popular methods in brain imaging field. First method, dynamic expectation maximization (DEM) is a routine algorithm of inverting hemodynamic model. Second method, square root cubature kalman filtering and smoothing (SCKF-SCKS) based on the cubature kalman filter (CKF). Third method, the recently introduced confounds SCKF-SCKS (CSCKF-CSCKS) is modified model of the second method and worked at low interference factors. In this paper, the improvement of CSCKS algorithm are provided due to some major problems in traditional algorithm and obviously removes problems to get better results. The improved CSCKS algorithm estimates hemodynamic model parameters that are 89% close to real value, whereas the traditional CSCKS and DEM algorithms estimate parameters that are 82% and 53% close to real value, respectively. To compare the BOLD responses, a new parameter named power spectral density (PSD) is measured in this research work, which shows that the new CSCKS method produce the minimum BOLD signal strength 10-11dB at 3220Hz frequency, whereas SCKS and DEM methods shows 10-14dB at 3020 Hz and 10-16 dB at 2980 Hz frequency, respectively. This proves the ability of improved CSCKS method is to solve the hemodynamic model perfectly than that of others.
Keywords: Cubature kalman filter; blood oxygen level dependent signal; hemodynamic model; functional magnetic resonance imaging.
May 26 2024 , 11:38 AM
For ost recent news you haave to pay a visit internet and on internet I found this web site as a best site for hottest updates. https://www.waste-Ndc.pro/community/profile/tressa79906983/
28-Feb-2025
