Image Analysis

In patients, T2 hyperintense lesions were manually contoured on T2-weighted images with reference to fluid-attenuated inversion recovery images. T1 spin-echo hypointense and gadolinium-enhancing lesions were, respectively, contoured on pre- and post-gadolinium T1-weighted spin-echo sequences. The corresponding lesion masks were generated and aligned to the individual T1-weighted image using FLIRT (FMRIB’s [Oxford Centre for Functional MRI of the Brain’s] Linear Image Registration Tool) (FMRIB Software Library, version 5.0.9). After lesion-filling in patients only (21), T1-weighted scans were segmented using FreeSurfer, version 6.0.

The following regions of interest (ROIs) were then defined in patients: NAWM, defined as the WM outside visible lesions on T2-weighted images; perilesional area, defined as the 2-voxel rim outside lesional borders; T2 lesions; T1 spin-echo lesions; and gadolinium-enhancing lesions. In HCs, the WM ROI was defined.

In each subject, T1-weighted images were nonlinearly registered onto a standard brain image (Montreal Neurological Institute [MNI]) using a combination of affine and symmetric diffeomorphic transformations calculated with Advanced Normalization Tools, version 2.2. The derived transforms were used to align all previously generated ROIs to standard space.

Voxelwise ¹⁸F-DPA-714 distribution volume ratio (DVR) parametric maps were calculated using the Logan graphical method based on reference region, extracted using a supervised clustering algorithm (Supplemental Fig. 1; supplemental materials are available at http://jnm.snmjournals.org) (14). DVR maps were coregistered to the T1-weighted images with FLIRT and normalized to MNI using the previously derived transformations.