Mintun M, Clark C, Pontecorvo M, Lu M, Krautkramer M, Arora A, Joshi A, Veeraraj C, Skovronsky D.
Evaluation of a Binary Read Methodology for Florbetapir-PET Images.
Human Amyloid Imaging Abstract. 2012 Jan 1;
The report by Clark et al (JAMA,011) showed increased florbetapir uptake by visual analysis correlated to increased beta-amyloid deposition. To better define the procedure for determining whether a florbetapir-PET scan is normal or abnormal, a specific binary read methodology using visual analysis was defined and tested. This methodology specified an inverted gray scale, the manner of review of the images and the specific criteria for determination of a positive image. To evaluate readers using this methodology a series of example cases and practice cases were chosen and the training tested with cases in which the subject had come to autopsy and the neuropathology was known. For these autopsy cases a CERAD neuritic plaque count of ‘frequent’ or ‘moderate’ was considered positive, while “none” or “sparse” was considered negative. A total of9 nuclear medicine physicians were trained and evaluated.4 readers underwent in-person training while readers underwent “not-in-person” training via slides and recorded video. All readers had access to the training cases. All9 readers interpreted the5 autopsy cases from Phase III study reported by Clark et al and the median sensitivity was5% and specificity was4%. Their agreement with each other as measured by a Fleiss’ kappa was 0.82. Ten readers (5 with in-person and with not-in-person training) interpreted cases from an expanded set of9 autopsies yielding a median sensitivity of7% and specificity of5%. In this group, there were6 cases that came to autopsy within2 months. In this subset, the median sensitivity was1% and the specificity of4%.The readers with not-in-person training also interpreted an additional2 cases without autopsy validation that consisted of cognitively healthy controls, subjects with a diagnosis of probably AD, and those with MCI. The Fleiss’ kappa was 0.88 in these cases.