Artificial intelligence (AI) is changing how dementia is diagnosed. Artificial intelligence experts at Mayo Clinic have developed one approach focused on finding patterns in brain activity that may be hard to recognize with the human eye. AI tools can analyze brain scans and compare them to large databases to find patterns linked to specific types of dementia. This process can lead to earlier, faster and more-accurate diagnoses — even for people who don't have access to specialized medical centers and dementia experts.
A recent study tested StateViewer, an AI tool developed by researchers at Mayo Clinic to help diagnose dementia. The tool analyzes a type of brain scan called a fluorodeoxyglucose positron emission tomography (FDG-PET), which shows how different parts of the brain use sugar, also called glucose, for energy. Areas of the brain that are impacted by dementia tend to use less sugar. The FDG-PET scan highlights the areas where the brain uses less sugar, which can point to the type of dementia causing the symptoms. If the scan shows reduced sugar use in the areas of the brain involved in solving problems and remembering, for example, that can be an indicator of Alzheimer's disease.
StateViewer compares a scan to a large database of more than 3,600 FDG-PET scans from people already diagnosed with different types of dementia. By spotting patterns in brain activity and glucose usage, it can help identify different dementia types.
In the study, StateViewer correctly identified the dementia type 88% of the time. It also helped doctors reach a diagnosis faster than when using traditional methods. The tool is helpful in distinguishing between dementia types with similar symptoms, such as Lewy body dementia and posterior cortical atrophy.
StateViewer looks for patterns in brain activity and creates a color-coded brain map that provides a visual image of the affected areas in the brain. Healthcare professionals reviewing the scans can use these images to help diagnose types of dementia.
AI tools like StateViewer are designed to support — not replace — clinical decision-making. They offer a fast, consistent way to sort through complex information and give healthcare professionals more tools to help determine an accurate diagnosis.
The next steps are to determine whether AI tools may help detect dementia earlier, inform treatment plans and bring expert-level diagnostic support to more healthcare settings, including those without access to dementia specialists.