Abstract
PilotTones, proposed by Speier et al. are a new navigator method based on modulation of a locally generated low-power electromagnetic signal by physiological motion. The PilotTone (PT) signal is generated at a frequency close to the scanners Lamor frequency, inside the scanners receive-bandwidth but outside the MR signals bandwidth. The signal is then extracted in the reconstruction step. From this PT signal, a component can be extracted by using Independent Component Analysis (ICA) that represents cardiac motion. In this thesis, a processing chain is presented that enables retrospective evaluation of this cardiac component. Triggers obtained from this signal are evaluated against those from ECG in order to evaluate the feasibility of using the cardiac PT signal for prospective cardiac triggering. Furthermore, the FastICA algorithm (Hyvärinen et al.) is evaluated in terms of robustness and settings needed to reliably separate the cardiac signal.
Finally, the feasibility of real-time processing of the cardiac PT signal is demonstrated using an Extended Kalman Filter (EKF) to fit a phenomenological model based on the cardiac PT signal's frequency-content.
Finally, the feasibility of real-time processing of the cardiac PT signal is demonstrated using an Extended Kalman Filter (EKF) to fit a phenomenological model based on the cardiac PT signal's frequency-content.
Original language | English |
---|---|
Qualification | Master of Science |
Awarding Institution |
|
Supervisors/Advisors |
|
Award date | 25 Oct 2017 |
Publication status | Published - 2017 |
Externally published | Yes |
Keywords
- Cardiac MRI
- CMRI
- ECG
- Triggering
- ICA
- FastICA
- MRI
- Kalman Filter
- Extended Kalman Filter
- EKF