3D-DXA Cortical and Trabecular Parameters; Agreement and Precision Between GE Healthcare Prodigy and iDXA Densitometers

This study demonstrates that 3D-DXA cortical and trabecular parameters measured using GE Healthcare Prodigy and iDXA densitometers show excellent agreement and comparable short-term precision, indicating that no adjustments are necessary when using 3D-Shaper software across these different device models.

The Gist

Imagine your bones are like a complex, two-story house. For decades, doctors have used a special camera called a DXA scanner to take a flat, 2D "photograph" of this house to see how strong the walls are. This is the current gold standard for diagnosing osteoporosis (weak bones).

However, a new technology called 3D-DXA has arrived. Think of this as taking that flat photograph and using a smart computer program to build a virtual 3D model of the house. This allows doctors to see not just the outer walls (cortical bone) but also the internal scaffolding (trabecular bone), giving a much richer picture of bone health.

The problem? There are two different models of the "camera" taking the photos: the older Prodigy and the newer, sharper iDXA.

The Big Question

The researchers asked: If we take a photo of the same person with the old camera and the new camera, and then build a 3D model from both, will the models look the same?

Since the new camera (iDXA) takes higher-quality pictures, they worried the computer might build a slightly different 3D house, which could confuse doctors or require them to re-calculate everything when switching machines.

The Experiment

To find out, the team gathered 391 people from three different locations (Wisconsin, California, and Brazil). These people ranged from young athletes to older adults.

They put everyone through a "double-check" process:

1. They scanned people's hips with both the old and new cameras.
2. They used the 3D-Shaper software (the "architect") to build the 3D models from both sets of photos.
3. They compared the results to see if the two models matched up.

The Findings: A Perfect Match

The results were fantastic news for doctors and patients:

• The "Blueprints" Matched: Even though the cameras were different, the 3D models built from them were nearly identical. The correlation was so strong (over 96% agreement) that it's like taking a photo of a house with a vintage camera and a modern smartphone, and having the 3D reconstruction software produce the exact same blueprint.
• Tiny Differences Don't Matter: There were tiny, mathematically detectable differences between the two machines (like a difference of a few grains of sand in a pile of concrete). However, the researchers found these differences were clinically meaningless. They were so small that they wouldn't change a doctor's diagnosis or treatment plan. It's like measuring a room with a ruler that is off by a fraction of a millimeter; you still know exactly where the wall is.
• Consistency is King: The software was so good at its job that it produced consistent results even when the "photographer" (the technician) or the "camera" (the machine) changed. This is huge because it means the software can correct for small human errors in positioning, making the 3D models very reliable.

The Bottom Line

This study proves that you don't need to worry about which GE Healthcare scanner you use. Whether a patient is scanned on an older Prodigy machine or a newer iDXA machine, the 3D-Shaper software will give doctors the same accurate, reliable 3D picture of their bone health.

In simple terms: The new camera is sharper, but the smart software is so good at its job that it translates both the old and new photos into the same perfect 3D model. Doctors can switch between these machines without needing to recalibrate or adjust their treatment plans.

Technical Summary

1. Problem Statement

Dual-energy X-ray absorptiometry (DXA) is the clinical gold standard for diagnosing osteoporosis. Recent advancements, specifically 3D-DXA (using software like 3D-Shaper), allow for the reconstruction of standard 2D DXA hip scans into 3D models, enabling the measurement of distinct cortical and trabecular bone compartments (volumetric BMD and cortical surface BMD).

While the agreement and precision of standard areal BMD (aBMD) between different DXA scanner models (e.g., GE Healthcare Prodigy vs. iDXA) are well-documented, there is a lack of data regarding the interchangeability of 3D-DXA parameters across these different hardware generations. Given that the iDXA offers higher image quality than the older Prodigy, it was hypothesized that the reconstruction algorithms might yield different results, potentially requiring cross-calibration or adjustment factors when switching between devices.

2. Methodology

Study Design & Cohort:

• Participants: A total of 391 adults (men and women) recruited from three clinical centers: University of Wisconsin (USA), California State University, San Bernardino (USA), and Federal University of Rio de Janeiro (Brazil).
• Demographics: The cohort covered a wide age range (20–90 years) and diverse anthropometric profiles (BMI 17.4–48.8 kg/m²).
• Scanners: Subjects were scanned on either GE Healthcare Prodigy or iDXA densitometers.
  • University of Wisconsin (UW): Prodigy and iDXA.
  • CSUSB: iDXA.
  • FURJ (Brazil): Two Prodigy devices.
• Protocol: Subjects underwent duplicate hip scans with complete repositioning to assess short-term precision.

3D-DXA Analysis:

• Software: Analyses were performed using 3D-Shaper® software (v2.14).
• Method: The software registers a statistical 3D shape and density model onto the standard hip DXA scan to create a patient-specific 3D femur.
• Parameters Measured:
  1. Integral vBMD: Mean density of the total hip (cortical + trabecular).
  2. Cortical sBMD: Product of cortical volumetric BMD and cortical thickness.
  3. Trabecular vBMD: Average density of the trabecular compartment.

Statistical Analysis:

• Agreement: Assessed via Deming regression (assuming equal error variance) and Bland-Altman analysis (calculating bias and limits of agreement).
• Precision: Short-term precision was calculated using the Root-Mean-Square Standard Deviation (RMS-SD) and Root-Mean-Square Coefficient of Variation (RMS-CV) following International Society for Clinical Densitometry (ISCD) recommendations.
• Significance: A two-sided significance level of 0.05 was used.

3. Key Contributions

This study represents the first evaluation of inter-scanner variability for 3D-DXA parameters between GE Healthcare Prodigy and iDXA systems. It addresses a critical gap in clinical densitometry by determining if the transition from older (Prodigy) to newer (iDXA) hardware requires software adjustments for advanced 3D bone modeling.

4. Results

Inter-Scanner Agreement:

• Strong Correlation: There was excellent agreement between Prodigy and iDXA for all parameters (aBMD, Integral vBMD, Trabecular vBMD, Cortical sBMD), with R² values ranging from 0.96 to 0.99.
• Bias Analysis:
  • Statistically significant biases (p < 0.05) were detected in some centers for aBMD and 3D-DXA parameters.
  • Clinical Relevance: Despite statistical significance, the magnitude of these biases was clinically negligible. The biases were consistently less than one-third of the Least Significant Change (LSC) calculated for the devices.
  • Slope Differences: The absolute slope differences from unity were <6.0% for aBMD and <4.0% for 3D-DXA parameters, indicating strong linear agreement.
• Limits of Agreement: The intervals were similar across centers, suggesting that the differences between scanners are consistent regardless of the specific device model or location.

Short-Term Precision:

• aBMD Precision: Precision varied by center (CV% ≤ 1.12% at total hip, ≤ 2.46% at femoral neck), generally meeting or exceeding ISCD standards.
• 3D-DXA Precision:
  • Precision for 3D parameters was highly consistent across different scanners and centers.
  • RMS-SD Ranges:
    • Trabecular vBMD: 3.073 – 3.909 mg/cm³.
    • Cortical sBMD: 2.116 – 2.392 mg/cm².
    • Integral vBMD: 2.989 – 4.139 mg/cm³.
  • Comparison to aBMD: The precision of 3D-DXA measurements was more stable across different scanners than standard aBMD measurements. For example, aBMD precision at FURJ was ~66% lower than at UW, whereas 3D-DXA precision remained stable.

5. Significance and Conclusion

Clinical Implications:

• No Adjustment Required: The study concludes that no cross-calibration or adjustment factors are necessary when using 3D-Shaper software on either Prodigy or iDXA instruments. Clinicians can switch between these GE Healthcare devices without compromising the validity of 3D-DXA longitudinal monitoring.
• Robustness of 3D Modeling: The high consistency of 3D-DXA precision across different scanners suggests that the 3D-Shaper algorithm effectively corrects for variations in patient positioning and scanner-specific image quality. By estimating and correcting for femur orientation and automatically defining regions of interest on a personalized 3D mesh, the software minimizes operator-dependent variability.

Limitations:

• The study was limited to GE Healthcare devices (Prodigy and iDXA); results may not generalize to other manufacturers (e.g., Hologic).
• Precision assessments were primarily conducted in postmenopausal women, potentially limiting applicability to younger populations or men.
• The study was conducted in research settings, which may differ from routine clinical practice.

Final Verdict:
The study provides robust evidence that 3D-DXA parameters derived from GE Healthcare Prodigy and iDXA scanners are interchangeable with high agreement and comparable precision, supporting the seamless integration of 3D bone modeling into clinical workflows across different generations of DXA hardware.