Hessian matrix-based structure tensor analysis for fiber enhancement and direction encoding

摘要

Neuroanatomical connectivity is crucial for understanding brain functions and diseases. Quantitative analysis of fiber orientation and structural connectivity at single axon resolution has led to great challenges. To address this issue, Hessian matrix based structure tensor (HST) has been introduced for quantifying fiber architecture. Optimized parameters and block based processing pipeline has enabled whole brain fiber analysis with tunable scale from micro- to macroscopic level. The HST method has allowed for deciphering fiber density, color-encoded orientation and fiber tractography. Tested with Thy1-GFP mice datasets acquired with micro-optical sectioning tomography system, the proposed method enabled effective characterization of single axon/dendrite and ensemble of fibers in 3D space. HST has demonstrated the power for fiber quantification and the rationale for validation of diffusion tensor imaging.