RSAtrace3D

RSAtrace3D vectorizes root system architecture from 3D volume data, such as X-ray computed tomography (CT) images, to enable quantitative RSA phenotyping in monocot plants.

Key Features:

• Robust Vectorization: Converts complex root structures into polylines representing individual roots, including radicle and crown roots.
• Handling Noisy Data: Uses multiple base points along each root instead of relying solely on end nodes to improve vectorization in noisy 3D volumes.
• Tracking Algorithm: Determines vector direction using a tracking algorithm based on the center of gravity (COG) of root voxels.
• Data Compactness: Produces vectorized representations that are approximately 32,000-fold more compact than raw X-ray CT volumes.
• Versatility and Expandability: Demonstrated on rice (Oryza sativa) and extendable to other monocots and diverse 3D imaging modalities.
• Programming Language: Implemented in Python.

Scientific Applications:

• RSA phenotyping: Enables quantitative measurement and comparison of root system architecture traits in monocot plants.
• Varietal comparison and breeding: Facilitates detection of varietal differences in RSA traits relevant to breeding programs.
• Cross-modal analysis: Supports analysis of monocot root systems across different 3D imaging modalities.

Methodology:

Vectorization converts segmented root voxels into polylines using multiple base points per root and a tracking algorithm based on the center of gravity (COG) of root voxels to determine vector directions.