Feasibility study of microwave-induced thermoacoustic/ultrasound dual-modality imaging for the assessment of nonalcoholic fatty liver disease

Abstract

Background: Accurate and non-invasive diagnosis of nonalcoholic fatty liver disease (NAFLD) remains a critical challenge. Current methods are often costly, invasive, or lack sensitivity in early stages.

Purpose: To investigate the feasibility and accuracy of microwave-induced thermoacoustic/ultrasound (TAI/US) dual-modality imaging for the assessment of hepatic steatosis.

Methods: A microwave-induced TAI/US dual-modality imaging system with a microwave source at a center frequency of 3 GHz as an excitation source was applied to image rabbit NAFLD models, and the spleen-to-liver thermoacoustic amplitude ratio (TAR_S-L) parameter was introduced to assess hepatic steatosis. Pathology was used as the gold standard, and Oil Red O staining fat fraction (ORO-FF) was calculated by a semi-quantitative software. Group differences in TAR_S-L across hepatic steatosis grades were assessed using the Kruskal-Wallis test with Dunn-Bonferroni post hoc comparisons. The Mann-Whitney U test and Cliff's Delta (δ) were applied to compare dichotomized steatosis subgroups. Spearman's rank correlation coefficient was used to evaluate the association between TAR_S-L and ORO-FF. Receiver Operating Characteristic (ROC) curve was performed to assess the diagnostic performance of TAR_S-L, and the results were quantified by the Area Under the Curve (AUC). A p < 0.05 was considered significant.

Results: The results demonstrated that the thermoacoustic and ultrasound images from the microwave-induced TAI/US dual-modality imaging system could be well-fused. The median TAR_S-L values for normal (S0), mild (S1), moderate (S2), and severe (S3) steatosis were 1.110 (0.953-1.283), 1.260 (1.073-1.430), 1.740 (1.625-1.870), and 2.195 (1.833-2.580). TAR_S-L values increased with steatosis severity, showing a strong positive correlation with ORO-FF (r = 0.908, p < 0.001). Statistically significant differences with large effect sizes were observed in TAR_S-L between S0 vs. S3 and S1 vs. S3 (p < 0.05, |Cliff's δ| > 0.474), indicating marked group differences. While comparisons such as S0 vs. S2 and S1 vs. S2 lacked statistical significance (p > 0.05), large effect sizes (|Cliff's δ| > 0.474) suggest potential true differences limited by sample size. ROC analysis confirmed diagnostic performance, with AUC of 0.894 (0.764-1.000), 0.979 (0.931-1.000), and 0.953 (0.874-1.000) for distinguishing S0 vs. ≥S1, ≤S1 vs. ≥S2, and ≤S2 vs. S3, respectively.

Conclusion: At present, the diagnosis of NAFLD based on TAI still belongs to the research gap. Our study initially verified the potential of microwave-induced TAI/US dual-modality imaging combined with the thermoacoustic parameter TAR_S-L for the diagnosis of NAFLD. These findings establish a crucial methodological foundation for subsequent in-depth investigations into the application of microwave-induced TAI/US dual-modality imaging in the quantitative diagnosis of hepatic steatosis.

Keywords: microwave‐induced thermoacoustic/ultrasound dual‐modality imaging; nonalcoholic fatty liver disease; spleen‐to‐liver thermoacoustic amplitude ratio.