Dr. David Jones, creator of StateViewer and director of Mayo Clinic’s Neurology AI Program, analyzes brain scans with the AI tool that can identify nine types of dementia with 88% accuracy. Image from Mayo Clinic.
Every three seconds, someone in the world develops dementia, according to Alzheimer’s Disease International. But misdiagnoses and slow diagnoses are common problems as other diseases can mimic dementia. Added to that is the fear and stigma surrounding cognitive decline often keeps people from seeking help until it’s too late. Now, Mayo Clinic researchers have developed an AI tool that could alter this landscape by identifying nine different types of dementia, including Alzheimer’s disease, with 88% accuracy using just a single brain scan. The research, published in Neurology, details a new framework called StateViewer that draws on fluorodeoxyglucose PET (FDG-PET) imaging. In a study involving radiologists, the tool enabled clinicians to make correct diagnoses 3.3 times more often than with standard workflows, while interpreting brain scans nearly twice as fast.
Dr. David Jones, a Mayo Clinic neurologist and director of the Mayo Clinic Neurology Artificial Intelligence Program, emphasized how the brain’s complexity translates into diagnostic challenges when patients present with overlapping symptoms across different dementia types. StateViewer was designed to address this variability while maintaining focus on individual patient needs. “As we were designing StateViewer, we never lost sight of the fact that behind every data point and brain scan was a person facing a difficult diagnosis and urgent questions,” says Dr. Leland Barnard, the data scientist who led the AI engineering. The tool’s ability to detect patterns across nine dementia types addresses a fundamental problem: even experienced specialists struggle to distinguish conditions like Alzheimer’s from Lewy body dementia or frontotemporal dementia, particularly when multiple pathologies coexist.
Other researchers have also explored machine learning techniques to better understand dementia. For instance, Boston University’s team published work in Nature Medicine in 2024 using multimodal data from over 51,000 participants to differentiate 10 dementia types, while Cambridge University developed an algorithm that can predict Alzheimer’s progression from a single MRI scan with over 80% accuracy. What distinguishes Mayo’s StateViewer is its focus on FDG-PET imaging—a scan that reveals how the brain uses glucose for energy—combined with a visual interface that shows clinicians exactly why the AI reached its conclusion through color-coded brain maps. This transparency could be crucial for clinical adoption, as physicians can verify the AI’s reasoning against their own expertise rather than trusting a “black box” algorithm.
