An Automated Technique for Statistical Characterization of Brain Tissues in Magnetic Resonance Imaging

| A procedure for estimating the joint probability density function (pdf) of T 1 , T 2 , and proton spin density (P D) for gray matter (GM), white matter (WM), and cerebrospinal uid (CSF) in the brain is presented. The pdf's have numerous applications, including the study of tissue parameter variability in pathology and across populations. The procedure requires a multispectral, spin echo magnetic resonance imaging (MRI) data set of the brain. It consists of ve automated steps: 1) preprocess the data to remove extracranial tissue using a sequence of image processing operators; 2) estimate T 1 , T 2 , and P D by tting the preprocessed data to an imaging equation; 3) perform a fuzzy c-means clustering on the same preprocessed data to obtain a spatial map representing the membership value of the three tissue classes at each pixel location; 4) reject estimates which are not from pure tissue or have poor ts in the parameter estimation, and classify remaining estimates as either GM, WM, or CSF; 5) compute statistics on the classiied estimates to obtain a probability mass function and a Gaussian joint pdf of the tissue parameters for each tissue class. Some preliminary results are shown comparing computed pdf's of young, elderly, and Alzheimer's subjects. Two brief examples applying the joint pdf to pulse sequence optimization and generation of computational phantoms are also provided. 2 Removal of extracranial tissue: (a) P D-weighted image, (b) maximum membership segmentation, (c) class with highest valued centroid, (d) after morphological opening, (e) after island removal and region growing, (f) after hole removal and closing.. .. 26 3 Data set after preprocessing: (a) TR/TE=3000/30 ms, (b) TR/TE=3000/60 ms, 4 Parameter estimates: (a) T 1 Estimate, (b) T 2 Estimate, (c) D 0 Estimate, (d) q-6 Tissue classiication and thresholding operations: (a) Maximum membership seg-mentation into GM(medium gray), WM(light gray), and CSF(dark gray), (b) after membership thresholding, (c) after membership and q-value 7 Histograms and Gaussian joint pdf's of GM(medium gray), WM(light gray), and CSF(dark gray): (a) histograms at 0:3 max isosurface, (b) Gaussian pdf's at 0:3 max isosurface, (c) GM and WM histograms at 0:2 max magniied, (d) GM and 8 Comparison of GM(medium gray) and WM(light gray) joint pdf's: (a)-(e) young subjects, (f)-(h) elderly subjects, 9 Visualization of SPGR pulse sequence vs. GM(medium gray), WM(light gray), and CSF(dark gray) joint pdf's: (a) SPGR isosurface of 115, pdf isosurface of …