Grant: Catalyst Grant: Secondary Analysis of Neuroimaging Databases

When a patient first shows signs of possibly having a form of dementia, even experienced physicians using the best diagnostic tools currently available have difficulty in accurately determining what type of dementia that patient might have or even if the signs are simply those of normal aging. This is unfortunate, as treatments for one type of dementia may not be effective for another type and treatments may be of most benefit if given to patients early in their disease progression. The intent of this proposal is to develop methods, based on computerized image analysis of medical images, to diagnose different types of dementia earlier and more accurately than could otherwise be done. The methods will synthesize information from several types of images, including types that show details of brain anatomy (such as Magnetic Resonance Imaging, MRI) and types that show variations in brain physiology (such as Positron Emission Tomography, PET, or Single Photon Emission Computed Tomography, SPECT). To derive the methods, we will make use of data collected during the large, multi-centre, Alzheimer's Disease Neuroimaging Initiative (ADNI) study. This and similar large databases contain patient images and information about the patients' dementias that we can use to train an automated diagnostic system. This system should aid physicians in determining the most appropriate course of treatment for patients with dementia and should also help in the development of new treatments by helping in the selection of patients for clinical trials.