DIFFERENT TIME COURSE OF ASTROCYTOSIS AND AMYLOID DEPOSITION IN ALZHEIMER'S APPSWE TRANSGENICE MICE: A MULTI-TRACER MICROPET STUDY

Background: Texture analysis is an emerging method to describe variations in image intensity or pattern, which is undetectable to the eye. We investigated MR texture-analysis (MRTA) as a novel biomarker for detecting neurofibrillary tangles (NFTs) in tau specific animal model (rTG4510) of Alzheimer’s disease (AD). Texture analysis has been successfully applied to the classification of neurological abnormal tissue in human AD[1-3]. To evaluate MRTA as a clinical biomarker of pathology requires the use of animal models where in-vivo images can be correlated with histology in the same animal for the same period of incubation of AD. We applied TexRAD[4], an MRTA pipeline, which has previously demonstrated sensitivity to neurological disorders and correlations with adverse biology[5-8], to extract textural features based on image filtration-histogram and correlated this to immunohistochemistry (PG-5) findings of NFT’s[9]. Methods: Invivo experiments were performed on a 9.4T Agilent horizontal bore scanner. RF transmission was performed with a 72mm volume coil and a 4 channel receiver coil was used to image 9 transgenic and 16 wild-type 8.5 month old litter matched mice. The animals were imaged at 1.5% isofluorine and 1 L/mO 2. Body temperature was maintained at 37 o C and physiological monitored for temperature and respiration. AT2 weighted, 3D fast spin-echo sequence was implemented FOV1⁄419.2mmx16.8mmx12.0mm; resolution1⁄4150mmx150mmx150mm; TR1⁄42500ms, TEeff1⁄443ms, ETL1⁄44; NSA1⁄41. The scans were imported into TexRAD and 2D ROI’s were manually drawn in the cortex, hippocampus and thalamus. MRTAwas applied in two steps 1) filtration extract and enhances features of different sizes-fine, medium and coarse textures followed by 2) texture quantification using a histogram analysis-mean, proportion of positive pixels(PPP), entropy, kurtosis, skewness and standarddeviation. Results: Texture parameters were markedly different between wild type and TG4510 in the cortex (mean, p1⁄40.022, PPP, p1⁄40.007, entropy, p<0.001, kurtosis, p1⁄40.022), hippocampus (mean, p1⁄40.003, PPP, p1⁄40.006, entropy, p<0.001, skewness, p1⁄40.011) and in the thalamus (entropy, p<0.001, kurtosis, p1⁄40.034, skewness, p1⁄40.025). We observed good correlation between histology and kurtosis in the cortex, hippocampus and thalamus(figure 1) . Conclusions: MRTA successfully differentiated wild-type and tg4510 textural changes in regions of the brain susceptible to AD and the kurtosis measurement correlated with histological measures of NFT’s in each of these regions. Further work is underway to demonstrate if kurtosis maybe a clinically significant marker of AD at earlier time points.