Keynote

Multi-Scale Mechanobiological Simulations for Predicting and Modifying Femoral Growth in Children

  • Kainz, Hans (University of Vienna)
  • Egner, Clara (University of Vienna)
  • Kitir, Kevin (University of Vienna)
  • Rathmair, Laura (University of Vienna)
  • Mindler, Gabriel (Orthopedic Hospital Speising)
  • Kranzl, Andreas (Orthopedic Hospital Speising)
  • Koller, Willi (University of Vienna)

Please login to view abstract download link

Bones are essential for long-term health, providing structural support and enabling mobility. However, several patient groups develop torsional bony deformities during growth, impairing function and leading to long-term clinical complications. Understanding, monitoring, and modifying bone growth is therefore critical for improving mobility and preserving lifelong health. Over the past decade, mechanobiological models of bone growth have been used to investigate in vivo growth patterns in typically developing and pathological populations [1-5]. These multi-scale simulations integrate gait analysis, medical imaging, musculoskeletal modeling, and finite element (FE) analysis to estimate tissue-level loads and predict growth trends. However, clinical translation has been limited by the time-intensive creation of subject-specific models restricting studies to small sample sizes, and by the lack of longitudinal data for validation. To address these challenges, we developed streamlined workflows enabling large-scale, subject-specific simulations [6,7]. In parallel, we collected a unique longitudinal dataset comprising MRI and 3D motion capture data from 50 growing children to calibrate model parameters and evaluate predictive accuracy [8]. This presentation summarizes three recent advances: (i) a fully automated pipeline for medical image segmentation and femoral geometry quantification; (ii) a robust framework for generating consistent musculoskeletal and FE models; and (iii) a proof-of-concept study assessing whether gait modifications can alter femoral growth plate loading and predicted growth trajectories. Automatic femoral segmentation using a pre-trained U-Net framework enabled reliable extraction of torsional measures essential for growth prediction. Subject-specific musculoskeletal models were generated by adapting a generic model to the segmented bones, facilitating rapid model creation needed for clinical translation. The proof-of-concept results indicate that in-toeing primarily influenced predicted femoral anteversion development, whereas out-toeing predominantly affected the neck–shaft angle. These findings suggest that targeted, personalized gait modifications may redirect femoral growth. If validated in larger cohorts, such non-invasive interventions could reduce the need for surgical correction.