Image-and-Label Conditioning Latent Diffusion Model: Synthesizing A \beta-PET From MRI for Detecting Amyloid Status

• Published in: IEEE journal of biomedical and health informatics Vol. 29; no. 2; pp. 1221 - 1231
• Main Authors: Ou, Zaixin, Pan, Yongsheng, Xie, Fang, Guo, Qihao, Shen, Dinggang
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.02.2025
• Subjects: A<named-content xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:ali="http://www.niso.org/schemas/ali/1.0/" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" content-type="math" xlink:type="simple"> <inline-formula> <tex-math notation="LaTeX"> beta</tex-math> </inline-formula> </named-content>-PET imputation; Aged; Alzheimer Disease - diagnostic imaging; Alzheimer Disease - metabolism; Alzheimer's disease; Amyloid beta-Peptides - analysis; Amyloid beta-Peptides - metabolism; Biological system modeling; Brain - diagnostic imaging; Brain - metabolism; Dementia; Diffusion model; Diffusion models; Diseases; Feature extraction; Female; Hazards; Humans; Image Interpretation, Computer-Assisted - methods; Image quality; Image synthesis; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Male; Medical diagnosis; Positron-Emission Tomography - methods
• ISSN: 2168-2194; 2168-2208; 2168-2208
• DOI: 10.1109/JBHI.2024.3492020

Deposition of <inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>-amyloid (A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>), which is generally observed by A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>-PET, is an important biomarker to evaluate subjects with early-onset dementia. However, acquisition of A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>-PET usually suffers from high expense and radiation hazards, making A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>-PET not commonly used as MRI. As A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>-PET scans are only used to determine whether A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula> deposition is positive or not, it is highly valuable to capture the underlying relationship between A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula> deposition and other neuroimages (i.e., MRI) and detect amyloid status based on other neuroimages to reduce necessity of acquiring A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>-PET. To this end, we propose an image-and-label conditioning latent diffusion model (IL-CLDM) to synthesize A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>-PET scans from MRI scans by enhancing critical shared information to finally achieve MRI-based A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula> classification. Specifically, two conditioning modules are introduced to enable IL-CLDM to implicitly learn joint image synthesis and diagnosis: 1) an image conditioning module, to extract meaningful features from source MRI scans to provide structural information, and 2) a label conditioning module, to guide the alignment of generated scans to the diagnosed label. Experiments on a clinical dataset of 510 subjects demonstrate that our proposed IL-CLDM achieves image quality superior to five widely used models, and our synthesized A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula>-PET scans (by IL-CLDM) can significantly help classification of A<inline-formula><tex-math notation="LaTeX">\beta</tex-math></inline-formula> as positive or negative.