Automated bone mineral density prediction and fracture risk assessment using plain radiographs via deep learning

Abstract

Dual-energy X-ray absorptiometry (DXA) is underutilized to measure bone mineral density (BMD) and evaluate fracture risk. We present an automated tool to identify fractures, predict BMD, and evaluate fracture risk using plain radiographs. The tool performance is evaluated on 5164 and 18175 patients with pelvis/lumbar spine radiographs and Hologic DXA. The model is well calibrated with minimal bias in the hip (slope = 0.982, calibration-in-the-large = -0.003) and the lumbar spine BMD (slope = 0.978, calibration-in-the-large = 0.003). The area under the precision-recall curve and accuracy are 0.89 and 91.7% for hip osteoporosis, 0.89 and 86.2% for spine osteoporosis, 0.83 and 95.0% for high 10-year major fracture risk, and 0.96 and 90.0% for high hip fracture risk. The tool classifies 5206 (84.8%) patients with 95% positive or negative predictive value for osteoporosis, compared to 3008 DXA conducted at the same study period. This automated tool may help identify high-risk patients for osteoporosis.

Fig. 1

Schematic representation of the workflow for hip and spine BMD estimation.

Fig. 2. The calibration plots for predicted-measured BMD.

The calibration plots show predicted BMD values against DXA-measured BMD values for assessment of model performance. a Five thousand one hundred and sixty-four pairs of predicted-measured hip BMD (5164 patients), and b 57,662 pairs of predicted-measured lumbar vertebral BMD (18,175 patients). Each point represents a data pair of predicted and measure BMD. The points close to the diagonal line suggests good calibration.