Deep Learning on Amyloid Positons Emission Tomography

Reducing injected dose and/or acquisition time in amyloid PET imaging would improve comfort, radiation safety and cost-effectiveness in diagnosis and follow-up of patients. This study evaluates the impact of a deep learning-based noise reduction algorithm on visual analysis and Centiloid quantification when simulating reduced injected doses of \[18F\]flutemetamol.

Inclusion Criteria: * Patients with objective cognitive impairment, * Referred to our department for a cerebral \[¹⁸F\]flutemetamol positron emission tomography scan between January 1, 2023 and July 1, 2025, Exclusion Criteria: * Patient have objected to the use of their data.