Microsoft Word - ISMRM2005-000535.DOC
نویسندگان
چکیده
In analysis of fMRI data, highly sensitive detection of activated voxels is needed. The high noise levels in typical fMRI data makes this impossible to achive without pooling data from several voxels before the signal detection. The most common approach, applied in for instance SPM [1], is to employ spatial lowpass filtering (using a fixed filter kernel) of the data prior to the detection of active voxels. This is done under the assumption that in most small neighborhoods, either almost all or almost no voxels are part of an activated region. Naturally, this causes blurring of the edges of activated areas, and depending on the chosen threshold either causes regions of detected activation to shrink or grow. Another method, proposed by Friman et al [2], uses canonical correlation analysis (CCA) to adaptively find a spatial lowpass filter that maximizes the similarity between the filtered data and the model of the blood oxygen level dependent (BOLD) signal. However, it is equally important to correctly classify inactive voxels, and maximizing the similarity in each neighborhood may cause voxels close to the boundary of an active region to be falsely declared as active.
منابع مشابه
Microsoft Word - ISMRM2005-003424.DOC
P. E. Larson, D. G. Nishimura, J. M. Pauly Electrical Engineering, Stanford University, Stanford, CA, United States Introduction: Ultra-short echo time (UTE) imaging has recently become of interest for possible clinical applications [1,2]. Long-T2 species dominate UTE images if they are not suppressed. We introduce a new set of long-T2 suppression pulses designed with the Shinnar-Le Roux (SLR) ...
متن کاملMicrosoft Word - ISMRM2005-002196.DOC
Y. Yang, H. Gu, E. A. Stein Neuroimaging Research Branch, National Institute on Drug Abuse, Baltimore, MD, United States Introduction Functional MRI (fMRI) techniques based on BOLD (blood oxygenation level-dependent) and CBF (cerebral blood flow) (1,2) contrasts have been used extensively for mapping functional neuroanatomy. Recently, an fMRI technique based on cerebral blood volume (CBV) chang...
متن کاملMicrosoft Word - ISMRM2005-004306.DOC
H. Park, R. Bhagalia, C. R. Meyer, B. Kim Department of Radiology, University of Michigan, Ann Arbor, MI, United States, Department of Electric Engineering and Computer Science, University of Michigan, Ann Arbor, MI, United States Introduction: Subject motion is a major source of localization error in fMRI studies. Generally used fMRI analysis methods employ volume alignment of EPI with the ass...
متن کاملMicrosoft Word - ISMRM2005-000291.DOC
B. Wu, P. Qu, J. Yuan, G. X. Shen MRI Lab, Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong INTRODUCTION Minimizing the mutual coupling among array elements is necessary for parallel MRI. Preamplifiers with extremely low input-impedance [1] and capacitive decoupling networks [2, 3] have become the primary techniques for decoupling arbitrarily placed co...
متن کامل