Search for: All records

Today’s generative models can synthesize magnetic resonance images (MRIs) of the brain at specific ages. However, such models can neither map the aging process longitudinally within subjects, nor accommodate its variability across subjects. Such approaches also cannot predict anatomic features of aging in ways that can be validated retrospectively or trusted prospectively. We introduce a three-dimensional hybrid ControlNet + diffusion model that uses the baseline T1-weighted MRIs of healthy adults to predict individual neuroanatomic aging trajectories, as reflected by follow-up MRIs. The approach captures individual anatomical changes with an average predicted voxelwise intensity error of 15% and structural similarity index of 93%. Unlike methods relying on qualitative validation, our approach quantifies the fidelity of prospective MRI synthesis using FreeSurfer volumetrics. Because brain atrophy reflects risk for Alzheimer’s disease (AD), our model’s ability to generate individual-specific prospective MRIs suggests its clinical potential to assist AD risk estimation. 

Computation graphs are Directed Acyclic Graphs (DAGs) where the nodes correspond to mathematical operations and are used widely as abstractions in optimizations of neural networks. The device placement problem aims to identify optimal allocations of those nodes to a set of (potentially heterogeneous) devices. Existing approaches rely on two types of architectures known as grouper-placer and encoder-placer, respectively. In this work, we bridge the gap between encoder-placer and grouper-placer techniques and propose a novel framework for the task of device placement, relying on smaller computation graphs extracted from the OpenVINO toolkit. The framework consists of five steps, including graph coarsening, node representation learning and policy optimization. It facilitates end-to-end training and takes into account the DAG nature of the computation graphs. We also propose a model variant, inspired by graph parsing networks and complex network analysis, enabling graph representation learning and jointed, personalized graph partitioning, using an unspecified number of groups. To train the entire framework, we use reinforcement learning using the execution time of the placement as a reward. We demonstrate the flexibility and effectiveness of our approach through multiple experiments with three benchmark models, namely Inception-V3, ResNet, and BERT. The robustness of the proposed framework is also highlighted through an ablation study. The suggested placements improve the inference speed for the benchmark models by up to over CPU execution and by up to compared to other commonly used baselines.