Alzheimer's Disease (AD) is a cortical dementia and is therefore well suited for a technique like Electroencephalography (EEG) which measures the cortical activity of the brain. The main EEG abnormalities in AD are slowed mean frequency and reduced coherence among cortical regions. The diagnostic accuracy of conventional EEG in AD is around 80%. Statistical Pattern Recognition (SPR) is a statistical analysis that uses a database of information, in this case features from EEG registration, to identify individuals with aberrant pattern of brain waves. By using this method, the intention is to increase the diagnostic accuracy of EEG in AD. The participants were 300 patients diagnosed with AD at a Memory Clinic and 400 normal individuals evenly distributed in the age range of 50 - 90 years. SPR was used to identify the patients using more than 600 features of the EEG registration. A numeric index from 0-100 was established to describe the likelihood of AD in each case. This method correctly identified AD patients when compared to the normal group with >90% accuracy. Furthermore, the AD-index correlated with the severity of the disease as evaluated by the MMSE. The method also correctly identified AD patients when compared to EEG in patients with vascular dementia (VD) with some overlap in patients with mixed VD and AD. By using SPR on features of EEG registration, this method is more sensitive in identifying patients with AD from normal individuals and from patients with VD than conventional EEG. The method is easy to use and is therefore more widely applicable than more specialized methods like MRI or SPECT/PET.
Read full abstract