This thesis addresses the problem of active noise control for fMRI systems. Reducing noise in an fMRI imaging space is important due to its effects on the patients' and technicians hearing, as well as activating spurious regions of the brain in fMRI research. The main contributions of this thesis includes designing the hardware and software setup for active noise control in a fMRI test-bed, modeling the primary and secondary paths of the fMRI test-bed for simulations, and implementing various active noise control algorithms in real-time. Based on observations from these real-time implementations, a novel hybrid algorithm is also developed that leverages the stability of the FxNLMS adaptive algorithm with the high convergence speed of the FxRLS adaptive algorithm. This work takes an important step towards reducing noise in an fMRI room by gathering relevant fMRI modeling data and developing a hybrid FxRLS-FxNLMS that takes practical algorithm implementation issues into consideration. |