A radiomics platform for computing imaging features from µCT images of Thoroughbred racehorse proximal sesamoid bones: Benchmark performance and evaluation.
Authors: Basran Parminder S, Gao Jonathan, Palmer Scott, Reesink Heidi L
Journal: Equine veterinary journal
Summary
# Editorial Summary: Radiomics Analysis of Proximal Sesamoid Bone Fractures in Racehorses Proximal sesamoid bone fractures remain the leading cause of fatal musculoskeletal injury in North American racehorses, yet detecting the structural changes that predispose bones to catastrophic failure has proven difficult using conventional imaging analysis. Basran and colleagues developed a radiomics platform to systematically extract and quantify 215 imaging features from micro-CT scans of proximal sesamoid bones, comparing specimens from eight horses with fracture history against eight unaffected controls (19 fractured and 30 intact limbs total). The analysis successfully validated established morphological markers—volume, minor axis dimensions and anisotropy—against previously published quantitative data, whilst additionally identifying significant differences (P < 0.01) in more subtle radiomics features such as entropy, coarseness and histogram characteristics that would be imperceptible to manual inspection. Notably, the sensitivity of these imperceptible features to different computational settings highlights the importance of standardised protocols when developing diagnostic thresholds. For practitioners involved in injury prevention, this work establishes a framework for identifying early structural vulnerabilities in sesamoid bones that might guide risk stratification of racehorses, though further research using in vivo imaging will be essential before clinical application.
Read the full abstract on PubMed
Practical Takeaways
- •Radiomics provides a standardized, quantitative approach to analyzing sesamoid bone CT images that could improve detection of fracture risk in racehorses, potentially helping identify at-risk animals before catastrophic injury occurs
- •The technique reveals subtle structural changes in bone that are not perceptible to visual inspection, offering a more objective diagnostic tool for equine practitioners managing high-value racing stock
- •Implementation requires standardized imaging protocols and software, but offers reproducible feature extraction that could support evidence-based risk assessment in racing medicine
Key Findings
- •A radiomics platform successfully computed 215 imaging features from μCT images of proximal sesamoid bones, with morphologic features (volume, minor axis dimensions, anisotropy) showing high correlation to previously published quantitative CT data
- •Radiomics analysis identified significant differences (P < 0.01) between fractured and control proximal sesamoid bones in both perceptible features and imperceptible features such as entropy, coarseness, and histogram parameters
- •Many more image feature differences were revealed through radiomics analysis compared to conventional quantitative CT approaches, demonstrating enhanced sensitivity for detecting structural changes