The detection and significance of subtle changes in mixed-signal brain lesions by serial MRI scan matching and spatial normalization

Fast, accurate, and reproducible automatic segmentation of the brain in T1-weighted volume MRI data

“…The main factors that may limit the applicability of our algorithm to abnormal brains are as follows: a) the presence of tissues in the brain with an intensity significantly lower than that of GM or greater than that of WM; and b) the presence of brain structures smaller The two datasets were registered using our software, MRreg (7,21); the resulting transformation matrix was applied to the extracted brain from the repeat scan and sinc-based interpolation (radius ϭ 7) used to reconstruct the matched scan. The segmented brain from the baseline scan was then subtracted from the registered segmented brain from the repeat scan after intensity matching.…”

“…Segmentation of the brain from MRI scans has important applications in neuroimaging, in particular for visualization and quantification of the shape of the cortex (1-4), as a preliminary step for analysis of the spatial distribution of grey matter (GM) (5), and for image registration (6,7).…”

“…For example, in the study of patients with medically intractable epilepsy, the quantification of signal changes in lesions (based on serial scanning and high-precision image registration) (7), the analysis of cortical morphology (4), the quantitative analysis of surgical resections (11), and correction of the partial volume effect in magnetic resonance spectroscopy (12) are among the main applications of this sequence. Currently, manual or semi-automated segmentation techniques are still widely used to extract the brain from the image data.…”

“…To make the algorithm independent of scan orientation and to ensure the full connectivity of all parts of the segmented brain (i.e., avoid errors due to lack of 2D connectivity), all operations are performed in 3D. The new algorithm is derived from an earlier 2D algorithm used for automatic brain extraction for the registration of serial MRI scans (7). Automatic foreground thresholding and in particular the ability to segment in 3D is greatly facilitated by the use of an efficient intensity non-uniformity correction (17).…”

Co‐registration of in vivo human MRI brain images to postmortem histological microscopic images

“…These similarity measures are based on image intensities, information measures, and object patterns, and have been described in detail by Hajnal [Hajnal et al 1995] and Holden [Holden et al 2000]. Specifically, the measures investigated here were: (1) Mean-Squared Difference (MSD) [Viola 1995, Kim et al 2000; (2) Pearson Cross Correlation (PCC) [Viola 1995, Lemieux et al 1998]; (3) Mutual Information (MI) [Maes et al 1997, Kim et al 1997, Kim et al 1999, Kjems et al 1999; (4) Normalized Mutual Information (NMI) [Studholme et al 1997, Studholme et al 2000a, Studholme et al 2000b]; (5) Entropy of Difference Image (EDI) ; (6) Ratio Image Uniformity (RIU) and Modified RIU (MRIU) [Woods et al 1998, Singh et al 1998, Holden et al 2000] and (7) Pattern Intensity (PI) ].…”