Technique for Suppression Random and Physiological Noise Components in Functional Magnetic Resonance Imaging Data

Magnetic Resonance Imaging (MRI) uses Radio waves and strong magnetic field rather than X-ray to provide clear and detailed pictures of internal organs and tissues. Functional Magnetic Resonance Imaging (fMRI) is a non-invasive brain imaging technique, which developed in the early 1990's, for determining which parts of the brain are activated by different types of physical sensation or activity, such as sight, sound, or the movement of subject's fingers, and detecting the corresponding increase in blood flow. Denoising procedure for functional Magnetic Resonance Imaging (fMRI) is introduced in this paper. The noises are classified into random noise components and physiological baseline fluctuation components. The proposed technique based on threshold the Fourier spectrum of the output response to remove any frequencies less than the fundamental frequency and harmonics of the true activation, which it is periodic. The proposed work was conducted using computer simulations data (block design), real baseline data with simulated activation patterns, as well as real data from event-related fMRI study on a normal human volunteer. The results show that, the new technique is suppressing both random and physiological noise components while preserving the true activation in the signal from the acquired data in a simple and efficient way. This allows the new method to overcome the limitation of previous techniques while maintaining a robust performance and suggests its value as a useful preprocessing step for fMRI data analysis.