Figure 2

Dynamic MRI data analysis. (a) Categorization of dMRI profiles: Typical data sets showing dynamic MRI signal intensity profiles categorized into non-enhancing and enhancing parent classes and high, medium and low ΔSI_(bolus) sub classes. Enhancing and non-enhancing pixels were identified according to whether curves returned to baseline (non-enhancing) or increased (enhancing) above baseline after the signal intensity reached its minimum value due to T2* losses. The extent of T2* induced signal loss determined the pixel sub-class and was, subsequently, assigned a colour. (b) Automatic cortex region of interest selection: Proton image (i) also shown with a brain mask (ii) calculated using thresholding and morphology. Morphological erosion of the brain mask (blue) yielded a cortex mask (iii) enabled an automatic, unbiased selection of a cortex region of interest (yellow). In addition, maps of the parent classes were obtained as shown in (iv) showing pixels that underwent contrast enhancement (blue) and those that failed to enhance (orange). (c) Cortical regions of interest: Cortex brain masks show the distribution of non-enhancing voxels with high (red), medium (orange) and low (yellow) ΔSI_(bolus) values. Green voxels, dominant in outer cortex layers from rats with meningitis, show enhancing voxels indicative of blood brain barrier breakdown.