TY - JOUR
T1 - Regression models for identifying noise sources in magnetic resonance images
AU - Zhu, Hongtu
AU - Li, Yimei
AU - Ibrahim, Joseph G.
AU - Shi, Xiaoyan
AU - An, Hongyu
AU - Chen, Yashen
AU - Gao, Wei
AU - Lin, Weili
AU - Rowe, Daniel B.
AU - Peterson, Bradley S.
N1 - Funding Information:
Hongtu Zhu is Associate Professor, Department of Biostatistics and Biomedical Research Imaging Center and Department of Radiology, University of North Carolina at Chapel Hill, NC 27599 (E-mail: [email protected]). Yimei Li is a Ph.D. student, Department of Biostatistics and Biomedical Research Imaging Center and Department of Radiology, University of North Carolina at Chapel Hill, NC 27599 (E-mail: [email protected]). Joshep G. Ibrahim is Alumni Distinguished Professor, Department of Biostatistics and Biomedical Research Imaging Center and Department of Radiology, University of North Carolina at Chapel Hill, NC 27599 (E-mail: [email protected]). Xiaoyan Shi is a Ph.D. student, Department of Biostatistics and Biomedical Research Imaging Center and Department of Radiology, University of North Carolina at Chapel Hill, NC 27599 (E-mail: [email protected]. Hongyu An is Research Assistant Professor, Department of Radiology, University of North Carolina at Chapel Hill, NC 27599 (E-mail: [email protected]). Yashen Chen is Research Fellow, Department of Radiology, University of North Carolina at Chapel Hill, NC 27599 (E-mail: [email protected]). Wei Gao is a Ph.D. student, Department of Radiology, University of North Carolina at Chapel Hill, NC 27599 (E-mail: [email protected]). Weili Lin is Professor, Department of Radiology, University of North Carolina at Chapel Hill, NC 27599 (E-mail: [email protected]). Daniel B. Rowe is Associate Professor, Department of Biophysics, Medical College of Wisconsin, Milwaudee, WI 53226 (E-mail: [email protected]). Bradley S. Peterson is Professor, Department of Psychiatry, Columbia Medical Center and the New York State Psychiatric Institiute, New York, NY 10032 (E-mail: [email protected]). This work was supported in part by NSF grants SES-06-43663 and BCS-08-26844 and NIH grants UL1-RR025747-01 and AG033387 to Dr. Zhu, NIDA grant DA017820 and NIMH grants MH068318 and K02-74677 to Dr. Peterson, NIH grants GM 70335 and CA 74015 to Dr. Ibrahim, and NIH grant R01NS055754 to Dr. Lin. The authors thank the editor, associate editor, and referees for their helpful comments and suggestions.
PY - 2009/6
Y1 - 2009/6
N2 - Stochastic noise, susceptibility artifacts, magnetic field and radiofrequency inhomogeneities, and other noise components in magnetic resonance images (MRIs) can introduce serious bias into any measurements made with those images.We formally introduce three regression models including a Rician regression model and two associated normal models to characterize stochastic noise in various magnetic resonance imaging modalities, including diffusion-weighted imaging (DWI) and functional MRI (fMRI). Estimation algorithms are introduced to maximize the likelihood function of the three regression models.We also develop a diagnostic procedure for systematically exploring MR images to identify noise components other than simple stochastic noise, and to detect discrepancies between the fitted regression models and MRI data. The diagnostic procedure includes goodness-of-fit statistics, measures of influence, and tools for graphical display. The goodnessoffit statistics can assess the key assumptions of the three regression models, whereas measures of influence can isolate outliers caused by certain noise components, including motion artifacts. The tools for graphical display permit graphical visualization of the values for the goodness-of-fit statistic and influence measures. Finally, we conduct simulation studies to evaluate performance of these methods, and we analyze a real dataset to illustrate how our diagnostic procedure localizes subtle image artifacts by detecting intravoxel variability that is not captured by the regression models.
AB - Stochastic noise, susceptibility artifacts, magnetic field and radiofrequency inhomogeneities, and other noise components in magnetic resonance images (MRIs) can introduce serious bias into any measurements made with those images.We formally introduce three regression models including a Rician regression model and two associated normal models to characterize stochastic noise in various magnetic resonance imaging modalities, including diffusion-weighted imaging (DWI) and functional MRI (fMRI). Estimation algorithms are introduced to maximize the likelihood function of the three regression models.We also develop a diagnostic procedure for systematically exploring MR images to identify noise components other than simple stochastic noise, and to detect discrepancies between the fitted regression models and MRI data. The diagnostic procedure includes goodness-of-fit statistics, measures of influence, and tools for graphical display. The goodnessoffit statistics can assess the key assumptions of the three regression models, whereas measures of influence can isolate outliers caused by certain noise components, including motion artifacts. The tools for graphical display permit graphical visualization of the values for the goodness-of-fit statistic and influence measures. Finally, we conduct simulation studies to evaluate performance of these methods, and we analyze a real dataset to illustrate how our diagnostic procedure localizes subtle image artifacts by detecting intravoxel variability that is not captured by the regression models.
KW - Diffusion tensor
KW - Goodness-of-fit statistic
KW - Influence measures
KW - Normal approximation
KW - Rician regression
KW - Visualization
UR - https://www.scopus.com/pages/publications/66549116371
U2 - 10.1198/jasa.2009.0029
DO - 10.1198/jasa.2009.0029
M3 - Article
C2 - 19890478
AN - SCOPUS:66549116371
SN - 0162-1459
VL - 104
SP - 623
EP - 637
JO - Journal of the American Statistical Association
JF - Journal of the American Statistical Association
IS - 486
ER -