Combining shear wave elastography and ultrasound localization microscopy for liver fibrosis staging in a rat model.
Authors
Affiliations (2)
Affiliations (2)
- National-Regional Key Technology Engineering Laboratory for Medical Ultrasound, Guangdong Key Laboratory of Biomedical Measurements and Ultrasound Imaging, School of Biomedical Engineering, Shenzhen University Medical School, Shenzhen, China.
- National-Regional Key Technology Engineering Laboratory for Medical Ultrasound, Guangdong Key Laboratory of Biomedical Measurements and Ultrasound Imaging, School of Biomedical Engineering, Shenzhen University Medical School, Shenzhen, China. [email protected].
Abstract
Accurate staging of liver fibrosis (LF) remains a clinical challenge. This study aimed to quantitatively evaluate the diagnostic value of combining shear wave elastography (SWE) and ultrasound localization microscopy (ULM) for LF staging across the full disease spectrum. We sought to integrate liver mechanical properties and microvascular structural parameters to build a multiparametric model for fibrosis staging. LF was induced in Sprague-Dawley rats and histologically staged according to the METAVIR system. A total of 62 rats were included in the final analysis. Both SWE and ULM were performed in each animal. Liver viscoelastic moduli were estimated from SWE data using SW-VEI-Net, a physics-informed neural network-based approach. ULM data were post-processed to reconstruct super-resolution images of the hepatic microvasculature, followed by quantitative analysis to extract microvascular structural parameters. Histology revealed fibrosis stages F0 (n = 10), F1 (n = 12), F2 (n = 19), F3 (n = 11), and F4 (n = 10). Compared with the single-modality SWE and ULM models, the multiparametric combined model achieved the highest numerical areas under the receiver operating characteristic curve across the four classification tasks: 0.93 for F0 versus F1-F4, 0.92 for F0-F1 versus F2-F4, 0.90 for F0-F2 versus F3-F4, and 0.89 for F0-F3 versus F4. Combining SWE-derived viscoelastic parameters with ULM-derived microvascular features enabled multiparametric characterization of LF in this preclinical rat model. Independent validation in larger animal cohorts and, ultimately, human studies is essential before robust clinical conclusions can be drawn. Question Can combining shear-wave elastography (SWE) and ULM improve the accuracy of LF staging? Findings By integrating viscoelastic and microvascular features, the SWE-ULM combined model achieved the highest numerical AUCs across all four fibrosis classifications in a rat model. Relevance statement The SWE and ULM combined model suggests potential value for LF staging by integrating viscoelastic and microvascular biomarkers, providing a basis for further investigation toward future clinical translation.