Search papers, labs, and topics across Lattice.
This paper introduces a knowledge distillation framework to predict amyloid-beta positivity from MRI scans alone, eliminating the need for PET imaging. A BiomedCLIP-based teacher model learns PET-MRI alignment using cross-modal attention and Centiloid-aware triplet contrastive learning, which is then distilled into an MRI-only student model via feature and logit matching. Experiments on OASIS-3 and ADNI datasets show effective knowledge transfer, achieving AUCs of 0.74 and 0.68 respectively, with saliency maps highlighting relevant cortical regions.
Ditch the expensive PET scan: this method predicts Alzheimer's amyloid-beta status directly from MRI with surprisingly good accuracy.
Detecting amyloid-$尾$ (A$尾$) positivity is crucial for early diagnosis of Alzheimer's disease but typically requires PET imaging, which is costly, invasive, and not widely accessible, limiting its use for population-level screening. We address this gap by proposing a PET-guided knowledge distillation framework that enables A$尾$ prediction from MRI alone, without requiring non-imaging clinical covariates or PET at inference. Our approach employs a BiomedCLIP-based teacher model that learns PET-MRI alignment via cross-modal attention and triplet contrastive learning with PET-informed (Centiloid-aware) online negative sampling. An MRI-only student then mimics the teacher via feature-level and logit-level distillation. Evaluated across four MRI contrasts (T1w, T2w, FLAIR, T2*) and two independent datasets, our approach demonstrates effective knowledge transfer (best AUC: 0.74 on OASIS-3, 0.68 on ADNI) while maintaining interpretability and eliminating the need for clinical variables. Saliency analysis confirms that predictions focus on anatomically relevant cortical regions, supporting the clinical viability of PET-free A$尾$ screening. Code is available at https://github.com/FrancescoChiumento/pet-guided-mri-amyloid-detection.
