Description

Magnetic Resonance Imaging (MRI) is a widely used medical imaging technique for noninvasive imaging of the human body. In addition to using MRI as a clinical diagnostic tool, functional MRI is used in biomedical research for functional brain mapping, or to find out which areas of brain are activated during certain tasks. Many secrets of the brain have been revealed during the past couple of decades thanks to fMRI. However, some of the current techniques and assumptions used in fMRI limit its potential as a research tool and its usefulness in clinical environment. The advancement of auditory related fMRI research is hindered due to systemic confound of acoustic noise associated with acquisition of every MRI image. This is confounded further by the fact that the auditory cortex, and the human brain in general, does not behave as a linear-time-invariant (LTI) system, which is a basic assumption in detecting activation in most of the fMRI studies. The nonlinearities observed in the auditory cortex cause an identical auditory stimulus to evoke slightly different hemodynamic responses (HDR) when delivered under different acoustic conditions, making activation detection harder and inaccurate. A novel approach to model the nonlinearities in HDRs was proposed and tested. The results showed that the response of primary auditory cortex can be characterized as the response of a state-dependent LTI system and the activation detection can be greatly improved using this new model.