Osteoporosis prediction from Frontal Lumbar Spine X-rays.

Area of Science:

• Radiology
• Medical Imaging
• Osteoporosis Research

Background:

• Osteoporosis diagnosis relies heavily on bone mineral density (BMD) measurements.
• Dual-energy X-ray absorptiometry (DXA) is the gold standard but can be inaccessible.
• Frontal lumbar spine X-rays (FLS-X) offer a potential alternative for BMD estimation.

Purpose of the Study:

• To evaluate the performance of DeepXray™ Spina software in estimating BMD and T-scores from FLS-X.
• To assess the software's accuracy in predicting osteoporosis classifications (normal, osteopenia, osteoporosis).

Main Methods:

• A Japanese cohort underwent both FLS-X and DXA within 30 days.
• DeepXray™ Spina estimated BMD from FLS-X; results were compared to DXA-measured BMD (observed BMD).
• Statistical analyses included Pearson's correlation coefficient (PCC) and normalized root mean square error (NRMSE); classification performance was assessed using accuracy, sensitivity, specificity, and Cohen's kappa.

Main Results:

• A strong correlation was found between estimated and observed BMD (PCC = 0.901, NRMSE = 0.070).
• The software achieved high accuracy in classifying osteoporosis (0.951) and osteopenia (0.854).
• Excellent performance metrics were reported for sensitivity and specificity across all classifications, with quadratic-weighted Cohen's kappa at 0.884.

Conclusions:

• DeepXray™ Spina demonstrates high performance and robustness in predicting osteoporosis from FLS-X.
• The software is a practical and reliable tool for osteoporosis assessment, potentially improving accessibility to diagnosis.