Search papers, labs, and topics across Lattice.
This paper introduces FlowLet, a novel conditional generative framework that synthesizes age-conditioned 3D brain MRIs using flow matching in an invertible wavelet domain. By addressing the limitations of existing generative methods, FlowLet reduces computational demands and avoids artifacts, enabling the generation of high-fidelity MRI volumes with fewer sampling steps. The approach significantly enhances brain age prediction performance, particularly for underrepresented age groups, while maintaining anatomical integrity in the generated images.
FlowLet can generate high-fidelity 3D brain MRIs with fewer sampling steps, enhancing brain age prediction for underrepresented demographics.
Brain Magnetic Resonance Imaging (MRI) plays a central role in studying neurological development, aging, and diseases. One key application is Brain Age Prediction (BAP), which estimates an individual's biological brain age from MRI data. Effective BAP models require large, diverse, and age-balanced datasets, whereas existing 3D MRI datasets are demographically skewed, limiting fairness and generalizability. Acquiring new data is costly and ethically constrained, motivating generative data augmentation. Current generative methods are often based on latent diffusion models, which operate in learned low dimensional latent spaces to address the memory demands of volumetric MRI data. However, these methods are typically slow at inference, may introduce artifacts due to latent compression, and are rarely conditioned on age, thereby affecting the BAP performance. In this work, we propose FlowLet, a conditional generative framework that synthesizes age-conditioned 3D MRIs by leveraging flow matching within an invertible 3D wavelet domain, helping to avoid reconstruction artifacts and reducing computational demands. Experiments show that FlowLet generates high-fidelity volumes with few sampling steps. Training BAP models with data generated by FlowLet improves performance for underrepresented age groups, and region-based analysis confirms preservation of anatomical structures.
