Premium
SU‐E‐I‐66: Quality Assurance Program for Functional MRI
Author(s) -
Damen N,
Zhou XJ
Publication year - 2013
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4814177
Subject(s) - imaging phantom , image quality , quality assurance , signal to noise ratio (imaging) , computer science , noise (video) , artificial intelligence , echo planar imaging , signal (programming language) , scanner , nuclear medicine , computer vision , physics , magnetic resonance imaging , optics , image (mathematics) , medicine , external quality assessment , pathology , radiology , programming language
Purpose: To develop and implement an automatic algorithm for quantitatively analyzing image quality parameters in functional MRI (fMRI) studies. The image quality parameters include signal stability over time, signal‐to‐noise ratio (SNR), and ghost levels. Methods: A set of images (9 slices, 512 images per slice over a time of 8′32′) in fMRI was acquired from a spherical quality assurance (QA) phantom using a gradient‐echo echo planar imaging (EPI) sequence at 3T. For a given slice, the center of phantom was determined automatically by calculating the center of the “mass”, followed by an automatic edge‐detection algorithm to outline the phantom edge. The signal intensity in the central region of the phantom was displayed as a function of time to assess the signal stability. After a mask for the phantom was determined, the mask was shifted by 1/2 of the FOV along the phase‐encoding direction, producing a mask for the EPI ghost. The ghost mask was sub‐divided into four equal quadrants, and the average and maximum ghost intensities were evaluated in each quadrant. Ghost levels within the four quadrants were compared to infer specific root causes of the ghosts. Finally, a background noise region was automatically determined. The standard deviation and mean intensity in the noise region were used to evaluate the SNR and ghost percentage according to the ACR recommendations. Results: The above algorithm was successfully implemented in an open‐course JAVA script. The signal stability and ghost level in fMRI QA scans were automatically visualized without user intervention. Further, ghost characteristics in the four quadrants provided valuable information to reveal the causes of the ghosts, including B0‐eddy currents, linear eddy currents, and cross‐term eddy currents. Conclusion: We have developed and implemented an automated algorithm for QA analysis in fMRI studies. This algorithm is open‐source, and can be downloaded through a website.