New AI Technology Identifies Nine Dementia Types from Single Scan

Mayo Clinic researchers have introduced a groundbreaking artificial intelligence (AI) tool designed to significantly enhance the diagnosis of various forms of dementia. This innovative tool, named StateViewer, enables clinicians to detect nine distinct types of dementia, including Alzheimer's disease, from a single brain scan, marking a pivotal advancement in early and accurate dementia diagnosis.

According to a study published in Neurology, the official journal of the American Academy of Neurology, StateViewer achieved an impressive accuracy rate of 88% in identifying different dementia types. The AI tool also drastically reduces the time required for clinicians to interpret brain scans, providing insights nearly twice as fast and with up to three times greater accuracy compared to conventional methods.

This tool was developed after training and testing on over 3,600 brain scans, which included images from individuals diagnosed with various forms of dementia as well as those without cognitive impairments. The creation of StateViewer directly addresses a significant challenge in dementia care: the necessity for early and precise identification of the disease, especially when multiple conditions coexist. Timely and accurate diagnosis is crucial for ensuring that patients receive the most suitable treatments when these can be most effective.

Dementia currently impacts more than 55 million individuals globally, with approximately 10 million new cases arising each year. Alzheimer's disease, the most prevalent type, is recognized as the fifth leading cause of death worldwide. The traditional diagnostic process for dementia is complex and often requires a combination of cognitive tests, blood tests, imaging, clinical interviews, and referrals to specialists. Even seasoned professionals frequently encounter difficulties in distinguishing between conditions like Alzheimer's, Lewy body dementia, and frontotemporal dementia.

Under the leadership of Dr. David Jones, a neurologist at Mayo Clinic and head of the Neurology Artificial Intelligence Program, StateViewer was created to provide a more streamlined approach to these diagnostic challenges. Dr. Jones emphasizes the intricate nature of each patient's brain, which informs his dedication to achieving clearer diagnostic outcomes.

Collaborating closely with Dr. Jones, Dr. Leland Barnard, a data scientist and the lead engineer behind StateViewer, shares the vision of this tool's potential impact. He highlights that each data point and brain scan represents a person facing a difficult health journey, and the AI tool aims to provide physicians with timely and precise insights to better assist their patients.

The StateViewer tool operates by analyzing a specific type of brain scan known as fluorodeoxyglucose positron emission tomography (FDG-PET). This scan reveals how the brain utilizes glucose for energy. By comparing each scan against a vast database of scans from individuals with confirmed dementia diagnoses, StateViewer identifies unique patterns that correspond to specific types of dementia.

Different forms of dementia affect various areas of the brain: Alzheimer's disease typically impacts memory and processing regions, Lewy body dementia is associated with areas linked to attention and movement, while frontotemporal dementia alters regions responsible for language and behavior. StateViewer visually represents these patterns using color-coded maps, making it easier for clinicians, regardless of their neurology training, to understand the AI's findings and how they support the diagnosis.

Researchers at Mayo Clinic are optimistic about expanding the application of this tool and will continue to evaluate its effectiveness across diverse clinical environments, aiming to enhance dementia care globally.