SynthPET — Multimodal Deep Learning Framework for FDG-PET Synthesis from MRI in Alzheimer's Disease
ENTITYSynthPET is a multimodal deep learning framework that synthesizes FDG-PET (fluorodeoxyglucose positron emission tomography) images from T1-weighted MRI and arterial spin labeling (ASL) perfusion MRI data. The system was developed to address a significant barrier to Alzheimer's disease research and clinical care: FDG-PET is widely used for detecting metabolic changes associated with neurodegeneration, but it is costly and involves ionizing radiation exposure, limiting its accessibility and repeatability. By learning to synthesize FDG-PET images from MRI data — which is less expensive, more widely available, and does not involve radiation — SynthPET aims to make metabolic neuroimaging more accessible for AD diagnosis, monitoring, and research. The use of both T1-weighted structural MRI and ASL perfusion MRI as inputs leverages complementary information about brain structure and blood flow to approximate the metabolic information captured by FDG-PET. The multimodal input design reflects the insight that perfusion MRI (ASL) captures physiological information more closely related to metabolic activity than structural MRI alone, potentially improving the fidelity of synthesized PET images. The framework contributes to a growing body of work on AI-based synthetic medical imaging, where generative models learn cross-modality mappings to enable virtual imaging without the constraints of the target modality. Validation of SynthPET against real FDG-PET images in Alzheimer's disease…
Subscribe to read full wiki analysis
Full wiki content — including source provenance, contradiction tracking, related pages, and analyst notes — is available to Subscriber members.
View Plans & SubscribeNew users get a 7-day trial — no credit card required.