Cohen A, Bi W, Weissfeld L, Aizenstein H, McDade E, Mountz J, Nebes R, Saxton J, Snitz B, Price J.
Classification of Amyloid-Positivity in Controls: Comparison of Approaches.
Human Amyloid Imaging Abstract. 2012 Jan 1;
The focus of PiB-PET imaging studies is shifting increasingly towards detection of amyloid pathology in cognitively normal individuals, generating a need to define cutoffs for the earliest signs of amyloid deposition. We previously reported the iterative-outlier method (IO) using dynamic DVR data for generation of cutoffs in a group of2 controls. Here, we examine the generalizability of the IO method when extended to late-summed (SUVR) data and alternative subject cohorts and compare IO to sparse k-means clustering (SKM). Finally, we compare the objective methods to consensus visual reads.
62 cognitively normal controls were utilized elderly (72.9 ±.2yrs), aged5-55yrs. An alternative cohort was generated by adding data representative of2 very elderly (>85yrs) subjects. The objective IO and SKM methods were compared to each other and to visual reads by five experienced readers.
The IO and SKM methods identified an identical set of6 subjects (26%) as PiB(+) using DVR data and similar results were observed using SUVR summed0-70min post-injection. When SUVR(40-60min) data or the alternative cohort were used, the classification of these original2 controls by IO differed significantly from SKM and from the original IO classifications using DVR data. Readers agreed with SKM classifications in5/16 (94%) of PiB(+) and in0/46 (87%)of PiB(-) cases. Of the visually rated PiB(+) but PiB(-) by SKM, were classified as PiB(+) by SKM at-3yr follow-up. These data suggest the SKM method is much more generalizable than the IO method and will be a useful objective approach to classify subjects as amyloid positive. Visual reads may be more sensitive to early amyloid deposition, but this may reflect that this method visualizes voxels throughout brain, as compared to the region-of-interest based approach of the IO and SKM methods, applied thus far.