Predicting Scoliosis Induction and Correction via Spine Growth Modulation in Pigs and Humans
Tuesday, August 26, 2025
4:00 PM-6:00 PM
BIOMED PhD Research Proposal
Title:
Predicting Scoliosis Induction and Correction via Spine Growth Modulation in Pigs and Humans using Optimized Patient-Specific Finite Element Models and Deep Learning Methods
Speaker:
Christian D’Andrea, PhD Candidate
School of Biomedical Engineering, Science and Health Systems
Drexel University
Advisor:
Sriram Balasubramanian
Professor
School of Biomedical Engineering, Science and Health Systems
Drexel University
Details:
Adolescent idiopathic scoliosis (AIS) is a 3D spine deformity, defined by a lateral curvature and axial rotation that affects 1-3% of 10-16-year-olds. Treatment decisions are made based on predicting curve progression during growth. Finite element (FE) modeling has been applied previously toward this task; however, current models lack patient-specific prognostic factors such as skeletal maturity, limiting accuracy. The same limitations apply to FE models used to predict outcomes for anterior vertebral body tethering (AVBT), a surgical intervention for AIS that corrects curvature by modifying long-term vertebral growth. Current AVBT planning strategies have led to under- and over-correction, and implant failure remains difficult to predict; a predictive framework using patient-specific models and prognostic factors could improve efficacy and consistency of surgical outcomes.
In addition to AVBT, new growth modulation implants are being validated using preclinical porcine models. Such trials are time-consuming and expensive, so to reduce this burden, porcine spine FE models have been used to simulate induction and correction of scoliosis using tether implants. However, existing models use simplified vertebral geometry and human-derived material properties; incorporating detailed vertebral geometry and porcine-derived soft tissue properties could improve model accuracy and better characterize implant performance.
Using high-fidelity FE models intra-operatively to optimize surgical outcomes despite deviations from an initial plan is infeasible due to computational cost. Deep-learning-based approaches to rapidly obtain outputs equivalent to those generated by FE simulations have been reported. However, none of these approaches have produced FE-equivalent outputs for AIS progression or correction. Creating and validating such models could support robust surgical outcome prediction as well as other use cases such as rapidly exploring implant design and configuration spaces.
The overall objective of this work is to develop accurate and high-fidelity spine computational models to simulate spine growth and scoliosis induction/correction using growth-modulating implants, and to develop deep learning techniques to obtain near-instant simulation solutions to support improved surgical planning capabilities and accelerated implant development.
Contact Information
Natalia Broz
njb33@drexel.edu