ConTrack: finding the most likely pathways between brain regions using diffusion tractography.

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Biology

Background:

• Magnetic resonance diffusion-weighted imaging coupled with fiber tractography (DFT) is the sole non-invasive technique for mapping white matter pathways in vivo.
• Current DFT algorithms sometimes fail to identify known neural connections, limiting applications in areas like visual neuroscience.
• There is a need for improved DFT methods to accurately trace specific pathways between predefined brain regions.

Purpose of the Study:

• To develop a novel probabilistic diffusion tractography (DFT) algorithm, named ConTrack, to identify the most likely white matter pathways between two specified brain regions.
• To overcome the limitations of existing DFT algorithms in detecting known neural connections.

Main Methods:

• Developed ConTrack, a probabilistic DFT algorithm comprising three components: a pathway sampler, a pathway scoring algorithm, and an inferential step.
• Generated a large set of potential pathways using the sampler.
• Assessed pathway likelihood using the scoring algorithm and identified the most probable pathways connecting the regions.

Main Results:

• ConTrack successfully estimated known pathways in human data, with results consistent with high-quality deterministic algorithms.
• The algorithm identified valid, known pathways that were missed by other deterministic and probabilistic DFT methods.
• Separating the sampling and scoring steps in ConTrack proved crucial for its enhanced performance.

Conclusions:

• ConTrack offers a significant advancement in non-invasive white matter pathway mapping, particularly for identifying specific connections.
• The probabilistic approach of ConTrack enhances the sensitivity and accuracy of diffusion tractography.
• This algorithm has the potential to improve our understanding of brain connectivity in various neuroscience applications.