Kenichi Oishi1, Michelle M. Mielke2, Andreia V. Faria1, Michael I. Miller, Perer C. M. van Zijl3, Marilyn Albert4,5, Constantine G. Lyketsos2,5, Susumu Mori1,3
1Radiology, Johns Hopkins University, Baltimore, MD, USA; 2Psychiatry & Behavioral Sciences, Johns Hopkins University; 3Kennedy Krieger Institute; 4Neurology, Johns Hopkins University; 5The Johns Hopkins Alzheimers Disease Research Center
We have developed an image analysis tool in which information extracted from multiple MRI modalities, using disease-specific spatial filters, is combined to generate a disease score. This tool was tested as an automated method to predict the conversion from amnestic mild cognitive impairment (aMCI) to Alzheimers disease (AD). We created disease-specific filters for each modality and optimized the combination to separate AD from control participants, using a training dataset, and applied the tool to calculate disease scores of 22 aMCI patients. The disease score predicted the conversion better than a single-modality approach, indicating the potential value for clinical application.