G-RMOS: GPU-accelerated Riemannian Metric Optimization on Surfaces

• Published in: Computers in biology and medicine Vol. 150; p. 106167
• Main Authors: Jo, Jeong Won, Gahm, Jin Kyu
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.11.2022; Elsevier Limited
• Subjects: Acceleration; Algorithms; Alzheimer's disease; Atrophy; Brain mapping; Brain Mapping - methods; Cortex; Embedding; Embedding registration; GPU-acceleration; Graphics processing units; Gray Matter - diagnostic imaging; Hippocampus; Internal Medicine; Latency; Mapping; Neurodegenerative diseases; Neuroimaging; Optimization; Other; Pipelining (computers); Substantia grisea; Surface mapping; GPU-acceleration; Surface mapping; Embedding registration; Hippocampus; Cortex
• ISSN: 0010-4825; 1879-0534; 1879-0534
• DOI: 10.1016/j.compbiomed.2022.106167

Surface mapping is used in various brain imaging studies, such as for mapping gray matter atrophy patterns in Alzheimer’s disease. Riemannian metrics on surface (RMOS) is a state-of-the-art surface mapping algorithm that optimizes Riemannian metrics to establish one-to-one correspondences between surfaces in the Laplace–Beltrami embedding space. However, owing to the complex calculation with accurate one-to-one correspondences, RMOS registration takes a long time. In this study, we propose G-RMOS, a graphics processing unit (GPU)-accelerated RMOS registration pipeline that uses three GPU kernel design strategies: 1. using GPU computing capability with a batch scheme; 2. using the cache in the GPU block to minimize memory latency in register and shared memory; and 3. maximizing the effective number of instructions per GPU cycle using instruction level parallelism. Using the experimental results, we compare the acceleration speed of the G-RMOS framework with that of RMOS using hippocampus and cortical surfaces, and show that G-RMOS achieves a significant speedup in surface mapping. We also compare the memory requirements for cortical surface mapping and show that G-RMOS uses less memory than RMOS. [Display omitted] •Propose a GPU-accelerated parallel framework for fast surface mapping.•Design a batch scheme with a GPU kernel using instruction level parallelism.•Demonstrate improvement in computational time for brain surface mapping.