Automatic target validation based on neuroscientific literature mining for tractography.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Bioinformatics

Background:

• Tractography studies require extensive anatomical knowledge and manual literature review for target identification.
• Manual literature review is time-consuming and can be subjective.
• Automated methods are needed to supplement traditional literature reviews.

Purpose of the Study:

• To develop and evaluate text-mining models for automatically identifying potential brain targets for tractography.
• To compare the efficiency of text-mining against manual literature review for mapping brain connectivity.
• To assess the utility of text-mining for identifying targets across different species (rat, primate, human).

Main Methods:

• Applied text-mining models to neuroscientific literature (full-text articles and abstracts).
• Selected three brain structures (internal globus pallidus, subthalamic nucleus, nucleus accumbens) for analysis.
• Performed systematic literature reviews and probabilistic tractography for comparison.
• Evaluated text-mining performance using recall and precision metrics.

Main Results:

• Text-mining models identified three times more potential targets than manual curation.
• Achieved 98% recall and 36% precision, missing only one target compared to literature review.
• Demonstrated successful extraction of connectivity information from a large volume of publications.

Conclusions:

• Text-mining offers an efficient and comprehensive approach to identify brain targets for tractography.
• This tool can significantly aid neuroscientists in mapping brain connectivity.
• The developed text-mining tools are publicly available on the HBP Neuroinformatics Platform.