Large variations occur in bone density measurements of children when using different software

Area of Science:

• Paediatric Bone Health
• Medical Imaging Analysis
• Dual X-ray Absorptiometry (DXA)

Background:

• Paediatric dual X-ray absorptiometry (DXA) studies face technical challenges, including edge detection algorithm failures for the pediatric skeleton.
• Hologic offers alternative algorithms, including low-density software, to address these limitations in pediatric DXA analysis.

Purpose of the Study:

• To evaluate the effectiveness of low-density software for analyzing pediatric DXA scans.
• To compare the results obtained using pediatric-specific protocols with the standard adult protocol in children.

Main Methods:

• 103 children (aged 5-17 years) underwent lumbar spine and whole-body DXA scans on a Hologic QDR-4500W.
• Scans were analyzed using both standard adult protocols and experimental pediatric/low-density protocols.
• Bland-Altman analysis was used to assess agreement between the different analysis methods.

Main Results:

• Both lumbar spine and whole-body pediatric protocols yielded lower bone mineral density (BMD) results than adult protocols.
• The low-density lumbar spine protocol showed a mean difference of 9%, with failures in children under 11.
• The experimental whole-body pediatric protocol showed a mean difference of 11%, with discrepancies noted.

Conclusions:

• The pediatric algorithm is crucial for anteroposterior (AP) lumbar spine scans in children under 11 to prevent analysis failures.
• The adult algorithm demonstrated acceptable performance for whole-body scans in children aged 11 and older.
• Further improvements in whole-body density analysis are needed for younger children, and new algorithms require normative data generation.