METHOD TO REMOVE SPIKE NOISE IN MAGNETIC RESONANCE DATA USING OVER SAMPLING
Publication Date: 2013-Jan-08
The IP.com Prior Art Database
A technique for detecting and removing spike noise from magnetic resonance (MR) raw data, and typically, k-space data with oversampling is disclosed. The spike noise is detected by applying low-pass filter on k-space data directly. Amplitude difference or power difference is calculated between acquired data and filtered data. This difference is compared to a threshold acquired from power difference of white noise. In case power difference between acquired data and filtered data is greater than threshold, a spike noise is detected.
FIELD OF INVENTION
The present invention relates generally to data processing of magnetic resonance imaging (MRI) system, and more particularly to a method of removing spike noise in magnetic resonance (MR) raw data.
BACKGROUND OF THE INVENTION
Generally, magnetic resonance (MR) images are reconstructed from k-space raw data. K-space raw data gets corrupted with high amplitude value noise, resulting in gross striation artifact. Such noise is referred to as spike noise. The reason for generation of spikes is arcing or loose connection or other electrical/mechanical problems with the MR system. Spike noise is difficult to diagnose and fixing the spike noise issue usually consumes considerable time and expense.
Various conventional techniques are used for correcting striation artifacts generated during MRI. Typically, the striation artifacts are generated by corruption of single or multiple data points in the k-space data due to spike noise. One of the conventional techniques involves generating an image of a body area. The image is reconstructed from raw data collected in a spatial-frequency domain. As some of the data points in the raw data from which the image is generated are corrupted, the image of the body area contains striation artifacts. A region containing a homogenous material is selected. Subsequently, corrupted data points in the raw data are detected and corrected according to Fourier transform correction process, and a new corrected image essentially free of striation artifacts is generated. However, the technique corrects the striation artifacts in the imaging object domain and not in k-space domain.
Another conventional technique describes detection and removal of random spike noise in MR raw data, and typically, k-space data by using a threshold. This threshold technique reduces or eliminates the corduroy-type and higher than usual level artifacts in MR images resulting from random spike noise in k-space data. The threshold technique involves applying a spatially varying threshold to be k-space data. Any data point that has a magnitude greater than that of the threshold value at that location is replaced by a local complex average of the neighboring data points or some other suitable data replacement scheme. However, the above mentioned technique isolates spike noise from threshold which is not very effective in detecting the spike noise in real time.
It would be desirable to have a technique which effectively detects and removes spike noise from MR raw data, and typically, k-space data.
BRIEF DESCRIPTION OF THE INVENTION
The invention discloses a technique to detect and remove spike noise from MR raw data, and typically, k-space data with oversampling. The spike noise is detected by applying low-pass filter on k-space data directly. Subsequently, power difference is calculated between acquired data and filtered data. The power difference is compared to...