Deriving pQCT-Type Spatial Density from DXA Images.

Area of Science:

• Biomedical Engineering
• Military Medicine
• Radiology

Background:

• Musculoskeletal overuse injuries are prevalent in military training, leading to significant costs and reduced readiness.
• Current injury risk models rely on peripheral computed tomography (pQCT), which is impractical for large-scale recruit screening.
• There is a need for accessible methods to assess injury risk in military personnel.

Purpose of the Study:

• To develop a method for deriving peripheral computed tomography (pQCT)-equivalent spatial density images from dual-energy X-ray absorptiometry (DXA) data.
• To assess the feasibility of using DXA as a practical tool for predicting musculoskeletal injury risk in military recruits.
• To establish a cost-effective screening protocol for stress fracture prevention.

Main Methods:

• Collected whole-body DXA and lower leg pQCT images from 51 male military recruits before training.
• Developed an artificial neural network (ANN) model to correlate DXA density profiles with pQCT-derived spatial bone density at specific tibial locations (38% and 66%).
• Validated the ANN model's predictions against actual pQCT data.

Main Results:

• The ANN model demonstrated a strong correlation (R² = 0.993 at 38% and R² = 0.990 at 66%) between predicted and raw pQCT spatial density.
• The derived pQCT images accurately represented bone density from lower-resolution DXA scans.
• The findings indicate high accuracy in predicting bone density using the developed method.

Conclusions:

• Dual-energy X-ray absorptiometry (DXA) can be practically utilized in place of peripheral computed tomography (pQCT) for assessing stress fracture risk.
• This approach holds high potential for mitigating musculoskeletal injuries in military recruits.
• A DXA-based protocol offers a feasible and accessible solution for injury risk assessment in military settings.