Non-invasive adjunctive diagnostic models incorporating high-frequency ultrasound and shear wave elastography for pediatric Duchenne muscular dystrophy

This is the authors' abstract. We don't add key points for this paper.

Background This study aimed to comprehensively characterize skeletal muscle properties in pediatric patients with Duchenne muscular dystrophy (DMD) using high-frequency ultrasound (HFUS) and shear-wave elastography (SWE), and to construct and compare diagnostic models based on these modalities to evaluate their potential as non-invasive adjunctive tools for assessing muscular involvement. Methods A total of 53 genetically or pathologically confirmed pediatric DMD patients and 61 age- and sex-matched healthy controls were enrolled. The bilateral biceps brachii and quadriceps femoris muscles were examined using HFUS and SWE. Inter-observer reliability was assessed using the intraclass correlation coefficient (ICC). Principal component analysis (PCA) was applied to HFUS parameters for dimensionality reduction, followed by logistic regression to establish a standardized HFUS diagnostic model. A separate SWE model was constructed using univariate and multivariate logistic regression. Receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA) were used to evaluate and compare the diagnostic performance of the individual and combined models. Results The HFUS model achieved an area under the curve (AUC) of 0.875 (95% CI: 0.807–0.943), with 88.9% sensitivity and 81.9% specificity. The SWE model showed an AUC of 0.846 (95% CI: 0.763–0.909), with 69.8% sensitivity and 75.4% specificity, with no significant difference between the two ( P = 0.494). Notably, the combined HFUS-SWE model yielded superior diagnostic accuracy (AUC = 0.947, 95% CI: 0.885–0.981; sensitivity = 82.2%, specificity = 95.1%), significantly outperforming each individual model (both P < 0.001). DCA demonstrated that the combined model provided a greater net benefit across all threshold probabilities. Conclusion Both the HFUS and SWE models demonstrated high diagnostic accuracy for DMD. Their combination significantly enhances diagnostic efficacy, offering a promising non-invasive adjunctive strategy for evaluating muscular involvement in children with DMD.