2020/06/19 by Davood Karimi, Lana Vasung, Karimi, Davood +11 · 1 citation
Medicine · #Advanced Neuroimaging Techniques and Applications #Bone and Joint Diseases #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #MRI in cancer diagnosis #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2006.11117
openalex publication_date 2020/06/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Multi-compartment modeling of diffusion-weighted magnetic resonance imaging\nmeasurements is necessary for accurate brain connectivity analysis. Existing\nmethods for estimating the number and orientations of fascicles in an imaging\nvoxel either depend on non-convex optimization techniques that are sensitive to\ninitialization and measurement noise, or are prone to predicting spurious\nfascicles. In this paper, we propose a machine learning-based technique that\ncan accurately estimate the number and orientations of fascicles in a voxel.\nOur method can be trained with either simulated or real diffusion-weighted\nimaging data. Our method estimates the angle to the closest fascicle for each\ndirection in a set of discrete directions uniformly spread on the unit sphere.\nThis information is then processed to extract the number and orientations of\nfascicles in a voxel. On realistic simulated phantom data with known ground\ntruth, our method predicts the number and orientations of crossing fascicles\nmore accurately than several existing methods. It also leads to more accurate\ntractography. On real data, our method is better than or compares favorably\nwith standard methods in terms of robustness to measurement down-sampling and\nalso in terms of expert quality assessment of tractography results.\n