Vitali Zagorodnov1, O. V. Ramana Murphy1
Most recent solutions to diagnose Alzheimers disease (AD) from structural brain measurements have been based on pattern classification framework, where the classifier score is used as a measure of disease progression. However, new classifier score are typically have to be learned for each new application, i.e. classification of AD patients vs. normal controls, prediction of MCI conversion or cognitive test scores. We derive a single universal application-independent classifier, which performs similar to or better than existing solutions that have been individually optimized to each of these applications.