How can machine learning help predict the clinical progression of Alzheimer’s?
Guilherme Bauer Negrini, Ph.D.
University of Pittsburgh
Pittsburgh, PA - United States
Alzheimer’s is characterized in part by the accumulation of an abnormal form of the tau protein into clumps called tangles. Tau tangles appear to form distinctive patterns in the brains of individuals with Alzheimer’s — patterns that differ from those seen in other forms of dementia. One technique for identifying such patterns in living individuals is via a brain scan called positron emission tomography (or PET). This imaging technique uses special “tracers” that highlight the amount and location of tau tangles in the living brain. However, the application of tau PET to clinical settings is limited by factors including the use of different tracers.
Dr. Guilherme Bauer Negrini and colleagues will use an advanced computer science technique called deep learning to overcome challenges associated with tau PET. The researchers will use tau PET data from more than 2000 individuals from six long-term studies to train their deep learning computer models. The team’s goal is to develop a freely available, user-friendly tool based on tau PET images that is generalizable across different tracers and may be used to predict the clinical progression of Alzheimer’s.
If successful, the results may contribute to the development of a tool that can aid in the assessment of Alzheimer’s in clinical settings and predict Alzheimer’s progression, regardless of the type of tau PET tracers or procedures used.
Back to Top