According to the International Diabetes Federation (IDF) 2025 report, the global number of diabetic patients is projected to exceed 700 million, with approximately 40% of type 2 diabetes mellitus (T2DM) patients developing diabetic nephropathy (DN). As the global incidence rate of diabetes continues to rise, the clinical diagnosis and treatment of DN have become increasingly critical. Although DN exhibits certain characteristic clinical manifestations, its early-stage symptoms often closely resemble those of non-diabetic renal diseases (NDRD), posing significant challenges to accurate diagnosis. Renal biopsy, as the gold standard for diagnosing DN, is limited in its widespread application due to its invasive nature. The innovative development and multimodal integration of ultrasound technology have increasingly highlighted its value in the differential diagnosis and disease assessment of DN. Conventional ultrasound techniques, including grayscale and Doppler ultrasound, evaluate renal morphology and hemodynamic changes. DN patients typically show increased kidney volume, enhanced renal cortical echogenicity, and elevated renal artery resistive index (RRI), which are closely associated with glomerular basement membrane thickening and reduced vascular compliance due to arteriosclerosis of the affe-rent arterioles. Ultrasound elastography provides a new dimension for assessing renal fibrosis by quantitatively measuring tissue stiffness. In DN patients, shear wave velocity (SWV) exhibits a characteristic pattern of "initial increase followed by decrease", which may correlate with histopathological staging. Contrast-enhanced ultrasound (CEUS) dynamically evaluates renal cortical microcirculation using microbubble tracking technology. CEUS images of DN patients demonstrate significantly reduced area under the curve (AUC) and peak intensity (PI), indicating decreased blood perfusion in the renal cortical microvascular bed. In recent years, the integration of artificial intelligence (AI) with ultrasound technology has advanced rapidly in the diagnosis and treatment of renal diseases. However, its deep integration with ultrasound for differential diagnosis and disease monitoring of DN has not yet been realized. In the future, combining AI algorithms with ultrasound technology is expected to enable automatic learning and identification of renal structures and pathological features from large volumes of ultrasound images, automatic quantification of key parameters such as RRI and SWV, and dynamic analysis of changes in renal microcirculation, thereby significantly improving the accuracy and efficiency of DN diagnosis.