JMIR Publications

ABSTRACT

Background:

Osteoporosis is a prevalent skeletal disorder characterized by decreased bone mass and increased fracture risk; however, it frequently remains underdiagnosed due to limited healthcare resources and its asymptomatic progression. Deep learning (DL) provides a promising solution for automated screening using computed tomography (CT) scans, enabling earlier detection and improved management.

Objective:

This systematic review and meta-analysis aimed to investigate the diagnostic performance of DL models in diagnosing osteoporosis based on CT scans.

Methods:

This study was conducted under the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines using articles extracted from PubMed, Scopus, Web of Science, and Embase. Quality Assessment of Diagnostic Accuracy Studies-2 was employed to estimate the risk of bias in each study. The confusion matrices from the included studies were extracted to summarize the diagnostic performance of DL models for osteoporosis.

Results:

This review included 18 studies, encompassing CT images from 21,759 participants for model development and validation. The pooled sensitivity and specificity were 0.90 (95% confidence interval [CI]: 0.86–0.93, I² = 79.02%) and 0.94 (95% CI: 0.91–0.96, I² = 88.97%) for osteoporosis diagnosis, 0.82 (95% CI: 0.77–0.86, I2 = 84.07%) and 0.92 (95% CI: 0.90–0.94, I2 = 79.80%) for osteopenia identification, and 0.92 (95% CI: 0.88–0.95, I2 = 98.02%) and 0.93 (95% CI: 0.90–0.96, I2 = 95.20%) for normal case identification. The area under the curve of the DL models for identifying osteoporosis, osteopenia, and normal cases were 0.97 (95% CI: 0.95–0.98), 0.94 (95% CI: 0.92–0.96), and 0.97 (95% CI: 0.96–0.98), respectively. Subgroup analyses revealed that models based on DenseNet variants, multislice input, CT as the reference standard, and three-dimensional architecture demonstrated superior diagnostic performance.

Conclusions:

This study supports the potential of DL models in diagnosing osteoporosis based on CT images. However, the results should be interpreted with caution due to the considerable heterogeneity among the studies. Further research is warranted to support their clinical translation and standardized application. Clinical Trial: PROSPERO (ID: CRD42024601713)