NerveTracker

NerveTracker visualizes and tracks groups of nerve fibers in serial block-face microscopy images using modified computer vision algorithms for microscopic tractography.

Key Features:

• Optic Flow Tractography: Modified optic flow algorithm tracks groups of peripheral nerve fibers along the length of the nerve.
• Structure Tensor Tractography: Modified structure tensor algorithm tracks groups of peripheral nerve fibers along the length of the nerve.
• Interactive Streamline Visualization: Provides interactive visualization of nerve fiber tracts.
• Manual Editing: Manual editing tools are provided for tractogram refinement.
• Deep Learning Fascicle Segmentation: Optional deep learning segmentation of fascicles constrains tracts from crossing into the epineurium.
• Robustness to Image Property Changes: Tractograms are robust to downsampling in-plane and out-of-plane with only 2% to 9% change to mean normalized Dice overlap values.

Scientific Applications:

• Peripheral Nerve Anatomy: Enables understanding of spatial organization of fibers within a nerve and its response to neuromodulation therapies.
• Fascicle Tracking: Identifies subsets of fibers from distinct fascicles that merge or split along the nerve's length, as demonstrated in vagus nerve samples where fibers from four distinct fascicles merge into a single fascicle over approximately 5 mm.
• Tibial Nerve Tractography: Performs tractography on tibial nerve datasets to assess accuracy of nerve fiber tracking.
• Accuracy Assessment: Compares resulting nerve tracts with segmentations of fascicles as they split and merge, achieving mean normalized Dice overlap above 0.75 across several millimeters.

Methodology:

NerveTracker modifies common computer vision algorithms (optic flow and structure tensor) to track groups of peripheral nerve fibers. Tractography accuracy is assessed by comparing resulting nerve tracts with segmentations of fascicles as they split and merge, using a normalized Dice overlap metric.