BoneDat, a database of standardized bone morphology for in silico analyses

Area of Science:

• Orthopedics
• Evolutionary Biology
• Medical Imaging
• Computational Biology

Background:

• In silico analysis of bone structure-function relationships is crucial but limited by the lack of standardized, high-quality human bone morphology datasets.
• Existing data limitations hinder research reproducibility and the development of reliable computational models for orthopedic and evolutionary studies.

Purpose of the Study:

• To develop BoneDat, a comprehensive database of standardized bone morphology data to overcome current research limitations.
• To provide curated, normalized bone geometry and reference morphology templates for enhanced computational modeling.

Main Methods:

• Collected and processed 278 clinical lumbopelvic CT scans (pelvis and lower spine).
• Included individuals aged 16-91, balanced by sex across ten age groups.
• Generated curated segmentation masks, normalized bone geometry (volumetric meshes), and reference morphology templates.

Main Results:

• BoneDat offers standardized reference geometry and enables shape normalization for improved computational model repeatability and credibility.
• The database facilitates the integration of other open datasets for training and benchmarking deep learning models.

Conclusions:

• BoneDat enhances the reliability and reproducibility of in silico bone analysis in orthopedics and evolutionary biology.
• The database accelerates the development and clinical application of deep learning models for bone morphology analysis.