[C-11]PIB PET scans as a measure of in vivo amyloid, and MRI scans for co-registration of the PET images, at baseline and eighteen months. Re- sults: Patterns of change in the individual measures (psychometrics/life- style information, PET images, and biomarkers) will be examined within each population group. Changes in psychometric measures will be corre- lated with neuroimaging and biomarker analysis to establish convergent validity of these measures. Conclusions: This will form the largest study of its kind ever undertaken in Australia. The current study will identify neuroimaging, biomarker and psychometric measurements of longitudinal changes in a large cohort, and will enhance knowledge of lifestyle and dietary factors associated with AD and healthy aging. P1-154 AUDITORY EVENT RELATED POTENTIALS: A CANDIDATE PHYSIOLOGIC BIOMARKER FOR EARLY DETECTION OF NEURODEGENERATION ASSOCIATED WITH ALZHEIMER DISEASE Christopher M. Clark 1 , Apostolos Topalis 2 , Deborah L. Green 3 , Nicholas Stepenosky 2 , Hardhik Gandhi 2 , Hannah McCoubrey 1 , John Kounios 3 , Sharon Xie 1 , Robi Polikar 2 . 1 University of Pennsylvania, Philadelphia, PA, USA; 2 Rowan University, Glassboro, NJ, USA; 3 Drexel University, Philadelphia, PA, USA. Contact e-mail: chris.clark@uphs.upenn.edu Background: Objective non-invasive methods to detect late-life neurode- generative dementia that can be obtained in community healthcare facilities will enhance the identification of individuals most likely to benefit from early intervention treatments. Event related potentials (ERP) reflect the integrity of neural networks and may provide a physiological biomarker of pathology. Non-stationary signal analysis with discrete wavelet transform (DWT) coupled to an automated feature classifier can be used in an EEG- based ERP assessment to aid in the early detection of neurodegenerative pathology. Objective: Using an auditory oddball ERP to assess the ability of DWT coupled with a multi-layered artificial neural network (ANN) to distinguish patients with very mild Alzheimer’s disease (AD) from cogni- tively normal elderly subjects. Methods: 23 AD (mean MMS 24, range 18-29) and 27 control subjects were evaluated. DWT decomposed scalp signals generated a time-frequency (multiresolution) wavelet analysis of the potentials. Data from the 1-2 Hz, 2-4 Hz, 4-8 HZ and 8-16 HZ bands were processed by an ensemble of multilayer perception classifier trained to distinguish AD patients from controls. Diagnostic statistics were calcu- lated using all available data. The performance of the classifier during the training phase was used to estimate the accuracy associated with the diagnostic decision generated for each subject. We also evaluated the ability of the ANN to distinguish mild (MMS 18 - 24) from very mild (MMS 24) AD using a stacked generalization algorithm followed by simple majority voting. Results: Ensemble ANN classification of DWT data from the 1-2 Hz, and 8-16 HZ bands provided a sensitivity of 79%, specificity 87%, positive predictive value of 87% and negative predictive value of 78% to distinguish AD patients from cognitively normal elderly subjects. An analysis of the training set data provided an estimated overall performance accuracy of 83%. The ANN was able to correctly classify the severity of the subject’s cognitive impairment with an overall accuracy of 76%. Conclusions: Oddball auditory ERPs linked to DWT and AAN analysis is a non-invasive physiological based biomarker with the potential to provide diagnostic and severity information in community healthcare facilities. P1-155 NEW DIAGNOSTIC MARKERS: THE EFFECTS OF SCOPOLAMINE ON EEG IN ALZHEIMERS DISEASE Jon Snaedal 1 , Kristinn Johnson 2 . 1 Landspitali University Hospital, Reykjavik, Iceland; 2 Mentis Cura, Reykjavik, Iceland. Contact e-mail: jsnaedal@landspitali.is Background: The diagnosis of Alzheimer's Disease (AD) is made mostly on clinical grounds and there is need for reliable biological markers for diagnosis in the prodromal phase of the disease. Electroencephalography (EEG) has been investigated as a possible surrogate biological marker for AD. A number of studies indicate that EEG is a sensitive measure of AD but not very specific. The aim of this study is to investigate how EEG can be made more clinically useful for early diagnosis of AD by using a cholinergic antagonist, scopolamine. Objective(s): Methods: We describe a study using advanced signal processing and pattern recognition algo- rithms together with administration of scopolamine which is a cholinergic muscarinic antagonist. Challenging the cholinergic system in the earliest phases is likely to reveal a response correlated to the disease as this system deteriorates significantly. We therefore hypothesize that the use of scopol- amine might enhance the sensitivity of EEG in AD. A clinical trial of 10 AD subjects, 10 patients with Mild Cognitive Impairment (MCI) and 10 age matched healthy individuals was conducted. A bank of 1000 EEG's is being established for future reference, the results of the first 200 will be presented. Results: A statistical pattern classifier was used to find which features best discriminate between AD and healthy controls. We found that features strongly correlated to the alpha generator (the source of the alpha activity) best discriminate between the groups. We obtained complete separation of the groups using our statistical pattern classifier approach which takes all features into account, increasing robustness of the method. The 10 MCI patients have now been followed for 18 months and 3 of them have progressed to AD. Their initial pattern was not distinguished from the pattern of the AD group. Conclusions: We demonstrate that this approach discriminates between the two groups and thus suggests that it may lead to improved early diagnosis of AD. with noninvasive methods. Furthermore, this approach may be helpful in discriminating persons with MCI progress- ing to AD from those who will not progress. P1-156 BRAZILIAN BRAIN BANK—2ND ANNIVERSARY— MORE THAN 1600 BRAINS COLLECTED Lea T. Grinberg, Jose Marcelo Farfel, Renata Eloah L. Ferretti, Renata E. Leite, Carlos Augusto Pasquallucci, Sergio Rosemberg, Ricardo Nitrini, Paulo Hilario Nascimento Saldiva, Wilson Jacob-Filho. University of Sao Paulo, Sao Paulo, Brazil. Contact e-mail: leagrinberg@usp.br Background: Brain banking remains a necessity for the study of brain aging physiology and pathology. The literature emphasizes the need for more postmortem surveys with elderly subjects, especially nondemented ones. The BABSGBB has been created in 2004 to provide material for the Brazilian Aging Brain Study Group. Objectives: 1) To carry on a large brain bank from people who died aged 50 years or over; 2) To facilitate neuroscience laboratories capable of studying normal aging brain processes and neurodegeneration; 3) To engage young researchers in Alzheimer’s Disease and related diseases’ studies. Methods: BABSGBB’s cases are gathered from autopsy cases of patients aged 50 years or over, sourced from the Sao Paulo Autopsy Service. Around 8,000 complete autopsies are performed per year in this age group. All protocols have been approved by the Local Ethics Committee and a written informed consent form is obtained from each participant’s relative. Brain parts are cryoprotected and fixated. CSF and blood are also collected. The neuropathological exami- nations are carried out based on widely accepted criteria, using immuno- histochemistry. The participant’s clinical and functional status are assessed based on the caregiver’s information, soon after the participant’s death, in order to gather antemortem cognitive, behavioral, and medical information. Results: From April 2004 to January 2006, 1,602 cases were gathered and classified as: CDR 0: 64.3%; CDR0.5: 12.8%, CDR1: 8.4%, CDR2: 5.2%, and CDR3: 9.3%. The mean age of participants is 68.2 years. Females are slightly more represented than males (54%:46%). Pure African descen- dents and Orientals comprise 11% and 1.3% of this series, respectively. On average, the cost for processing each case stood at US$200. To date, 13 Brazilian and 2 foreign groups are being directly benefited by this facility. In addition, 30 undergraduate and 6 graduate students have their primary training linked to the BABSG. Conclusions: The impact of this project is S141 Poster Presentations P1