Bone Mineral Density and Intermuscular Fat Derived from Computed Tomography Images Using Artificial Intelligence Are...

Area of Science:

• Radiology
• Orthopedics
• Artificial Intelligence

Background:

• Fracture healing assessment is crucial for patient outcomes.
• Current methods for evaluating fracture healing can be subjective and time-consuming.
• Objective quantitative markers are needed for early prediction of healing complications.

Purpose of the Study:

• To utilize artificial intelligence (AI) for automated measurement of bone mineral density (BMD) and paraspinal intramuscular fat area (PIFA) from computed tomography (CT) scans.
• To investigate the association between AI-derived BMD and PIFA with fracture healing outcomes in patients with rib fractures.

Main Methods:

• Retrospective analysis of CT scans from patients with rib fractures (2012-2023).
• AI-driven extraction of volumetric BMD (L1 vertebra) and PIFA (L1 midsection) from baseline CT scans.
• Logistic regression analysis to correlate BMD and PIFA with fracture healing outcomes (callus formation, volume increase, poor healing).

Main Results:

• A decrease in BMD was independently associated with reduced callus formation, diminished volume increase, and increased risk of poor healing.
• An increase in PIFA was independently associated with reduced callus formation, diminished volume increase, and increased risk of poor healing.
• A combined model of BMD, PIFA, and clinical factors significantly improved prediction of fracture healing outcomes compared to clinical factors alone (AUCs: 0.790, 0.749, 0.701).

Conclusions:

• Bone mineral density (BMD) and paraspinal intramuscular fat area (PIFA) measured by AI are significant early predictors of fracture healing.
• These AI-derived parameters can aid clinicians in identifying patients at risk for poor healing.
• Integrating BMD and PIFA into predictive models can enhance clinical decision-making for fracture management and intervention selection.