Glomerulonephritis through the lens of ultrasound and radiology

Abstract

Renal biopsy remains the diagnostic reference in glomerulonephritis, and no imaging modality has displaced its central role. The key question is whether imaging can complement histology by providing additional information on disease activity, prognosis, and treatment response. In clinical practice, conventional ultrasound is the first-line examination. Cortical thinning and elevated intrarenal resistive indices mainly reflect chronic structural damage and haemodynamic adaptation, without identifying the underlying pathological subtype. Techniques such as shear-wave elastography and contrast-enhanced ultrasound attempt to extend assessment beyond morphology by evaluating tissue stiffness and microvascular perfusion, with reported associations with fibrosis in immunoglobulin A nephropathy and microvascular changes in lupus nephritis and membranous nephropathy. Quantitative approaches integrating imaging features with clinical variables have also been explored for fibrosis staging, although broader validation remains limited. Computed tomography and magnetic resonance imaging continue to play a largely structural role, particularly in systemic diseases with renal involvement such as immunoglobulin G4-related nephropathy and vasculitis. Multiparametric magnetic resonance imaging can additionally assess perfusion and tissue microstructure through diffusion-weighted imaging, arterial spin labelling, and blood oxygenation level-dependent sequences. Nuclear medicine techniques may provide complementary molecular information, with exploratory applications in the assessment of inflammatory activity in antineutrophil cytoplasmic antibodies-associated vasculitis and in the evaluation of cortical or fibrotic involvement using newer tracers. Overall, evidence remains heterogeneous, and histological evaluation continues to determine diagnosis and guide treatment.

Key Words: Chronic kidney disease; Glomerulonephritis; Ultrasonography; Renal biopsy; Renal perfusion; Contrast-enhanced ultrasound; Lupus nephritis; Elastography; Magnetic resonance imaging; Renal fibrosis

Core Tip: Renal biopsy remains the gold standard for diagnosing glomerulonephritis, but imaging plays an increasingly important complementary role. Conventional ultrasound provides essential prognostic information, while advanced techniques, including contrast-enhanced ultrasound, elastography, multiparametric magnetic resonance imaging, and nuclear medicine, show promise for assessing fibrosis and disease activity. However, overlapping imaging findings across glomerulonephritis subtypes and lack of standardization limit their diagnostic specificity. These non-invasive modalities can aid in risk stratification and disease monitoring, but cannot yet replace biopsy for definitive diagnosis or therapeutic decision-making.

INTRODUCTION

Glomerulonephritis (GN) refers to a range of kidney diseases characterized by glomerular inflammation, which can progress to chronic kidney damage and ultimately end-stage renal disease. The global impact is considerable: Approximately 10%-15% of patients requiring renal replacement therapy in developed countries have primary glomerular diseases, with significant variation in incidence across regions and ethnic backgrounds[1]. Data from the Global Burden of Disease 2019 study demonstrate that worldwide cases of chronic kidney disease (CKD) due to GN rose from 9557400 in 1990 to 17308100 in 2019, representing a significant 77% rise over this three-decade period[2]. Notwithstanding advances in immunosuppressive therapies and supportive care, which have improved outcomes in recent years, many patients still risk irreversible kidney damage. This underlines the importance of early and accurate diagnosis, along with prompt therapeutic intervention[3].

Renal biopsy has traditionally been viewed as the definitive method for diagnosing GN, offering essential histopathological, immunofluorescence, and electron microscopic data that inform classification and treatment strategies[4]. However, the invasive nature of kidney biopsies, the associated risks of the procedure, and the challenges of obtaining multiple samples for disease monitoring have led to increased interest in exploring non-invasive alternatives[5]. In recent years, renal imaging has undergone a real transformation: From anatomical assessment to tools capable of investigating functional and molecular aspects of renal pathophysiology, providing real-time information about kidney pathophysiology[6]. The classic B-mode ultrasound continues to be the mainstay for examining kidney shape and the extent of structural modification correlated to the disease[7]. On the other hand, Doppler techniques provide important hemodynamic data via resistive index (RI) readings. Recent breakthroughs in ultrasound technology, such as contrast-enhanced imaging, elastography, and AI-enhanced radiomics, are now capable of distinguishing between inflammation and fibrosis, as well as differentiating various subtypes of GN[8-10].

Advanced imaging techniques have significantly broadened our diagnostic capabilities. Multiparametric magnetic resonance imaging (MRI) methods, such as diffusion-weighted imaging (DWI), arterial spin labelling, blood oxygenation level-dependent (BOLD), and T1/T2 mapping, enable assessment of tissue microstructure, blood flow, and fibrosis without invasive procedures. Additionally, new positron emission tomography (PET) radiotracers, including fibroblast activation protein inhibitor (FAPI) and Gallium-68 prostate-specific membrane antigen (⁶⁸Ga-PSMA), have shown promising predictive value for assessing the severity of fibrosis and information on cortical function[11-13]. Still, we face a fundamental challenge: GN is essentially a microscopic disease, involving immunological and cellular processes that current imaging systems cannot fully capture. Overlapping imaging phenotypes across different disease entities further complicate diagnosis. This review aims to present the current evidence on integrating multimodal imaging for GN management and to translate existing data into future prospects for GN diagnosis and management.

LITERATURE REVIEW

This narrative review was performed according to a structured and comprehensive literature search strategy. The electronic databases PubMed/MEDLINE, Scopus, and Web of Science were systematically queried from inception up to November 1, 2025. Search terms were selected to ensure high sensitivity and specificity and included controlled vocabulary (MeSH terms) and free-text keywords related to glomerular diseases and imaging modalities. The main search strategy combined terms such as “glomerulonephritis”, “renal imaging”, “ultrasonography”, “Doppler ultrasound”, “contrast-enhanced ultrasound”, “elastography”, “magnetic resonance imaging”, “positron emission tomography”, “radiomics”, and “artificial intelligence”, using Boolean operators. The search was restricted to peer-reviewed articles published in English. No a priori temporal limits were applied in order to capture both landmark and contemporary studies. Additional eligible records were identified through manual screening of reference lists of relevant articles. Eligible publications included original research articles, systematic reviews, meta-analyses, and high-quality narrative reviews addressing the application of imaging techniques in GN. Case reports and small case series were included only if they provided clinically relevant insights into emerging or innovative imaging technologies. Two authors independently screened titles and abstracts for relevance. Full texts were subsequently assessed for eligibility. Disagreements were resolved by consensus. The included studies were qualitatively synthesized and categorized according to imaging modality (conventional B-mode ultrasound, Doppler techniques, contrast-enhanced ultrasound (CEUS), elastography, multiparametric MRI, and nuclear medicine). Data extraction focused on technical principles, diagnostic performance, prognostic value, clinical applicability, and correlation with histopathological findings. Owing to the substantial heterogeneity in study design, patient populations, and outcome measures, a quantitative meta-analysis was not feasible; therefore, results were synthesized using a descriptive and comparative approach.

RENAL ULTRASOUND FINDINGS IN GLOMERULAR DISEASES

Conventional ultrasound, which combines B-mode imaging for structural evaluation and Doppler ultrasound for vascular assessment, is a central component of the initial diagnostic process and longitudinal follow-up for GN because of its safety, cost-effectiveness, bedside availability, and real-time assessment capabilities without ionising radiation[14-16]. As a non-invasive modality, ultrasound enables comprehensive evaluation of both structural features (such as kidney size, morphology, and parenchymal thickness) and functional parameters (including blood flow and vessel patency) in patients with abnormal kidney function[17]. It is particularly valuable in distinguishing reversible causes of acute kidney injury, such as obstruction, from parenchymal causes like GN[18], and provides essential real-time guidance for procedures such as renal biopsy, which is often required for definitive diagnosis[4,5,19]. The safety, non-invasiveness, and accessibility of ultrasound make it well-suited for longitudinal patient care in chronic conditions like GN[20]. While ultrasound offers significant advantages, its inherent non-specificity for distinct GN pathologies and limitations in detecting early-stage microscopic changes necessitate the continued essential role of renal biopsy for definitive diagnosis and precise histological staging[21]. However, ultrasound complements the clinical picture by providing real-time structural and hemodynamic information that enhances the diagnostic process, but does not replace the gold standard of tissue diagnosis[14].

Principles of conventional renal ultrasound

Conventional renal ultrasound primarily employs two main techniques: B-mode imaging for morphological assessment and Doppler ultrasound for evaluating vascular flow. These modalities provide complementary information crucial for comprehensive renal assessment[17,22].

B-mode imaging

B-mode ultrasound generates grayscale, cross-sectional images that assess kidney size, shape, parenchymal thickness, and echogenicity. In adults, a normal kidney measures approximately 10-12 cm in length and has a cortical thickness of 7-10 mm[6,22]. A kidney length less than 8 cm and significant cortical thinning with accentuated lobulation are typically signs of CKD[6]. Increased cortical echogenicity compared to the liver or spleen suggests parenchymal injury, correlating with fibrosis, glomerulosclerosis, and interstitial inflammation[15]. Although echogenicity assessment is subjective, changes over time can provide valuable clinical information. Quantitative studies have shown an inverse correlation between renal echogenicity and glomerular filtration rate (GFR), highlighting its potential for identifying irreversible advanced CKD[6].

Doppler ultrasound

Doppler ultrasound assesses renal perfusion and vessel patency[23] and is valuable for evaluating overall renal blood flow and detecting large-vessel abnormalities. Power Doppler increases sensitivity for detecting slow flow in small vessels, whereas spectral Doppler provides detailed waveform analysis for hemodynamic assessment. The renal RI (RRI), calculated as (peak systolic velocity - end diastolic velocity)/peak systolic velocity, reflects intrarenal vascular resistance and is associated with worsening renal function, higher chronicity index, and interstitial disease in GN. An elevated renal RRI may indicate impaired microcirculation and reduced renal function[24].

In advanced CKD, spectral Doppler waveforms typically show reduced diastolic flow and increased systolic velocities, reflecting interstitial fibrosis (IF) and glomerulosclerosis, especially when there is a loss of small blood vessels (microcirculatory rarefaction). A renal RRI ≥ 0.70 generally indicates tubulointerstitial damage and predicts progressive GFR decline. However, renal RRI elevation is more related to the degree of chronic injury than to specific GN subtypes, and may remain normal in early-stage primary GN despite reduced function[6].

Gigante et al[8] evaluated the RRI in 132 consecutive patients with biopsy-proven tubulointerstitial nephritis (TIN), comparing isolated-TIN vs GN associated TIN. Patients with isolated TIN had significantly higher renal RI values than those with TIN associated with non-immunoglobulin A nephropathy (IgAN) [0.73 (interquartile range = 0.68-0.77) vs 0.64 (0.60-0.67), P < 0.001]. This result confirms that the tubulointerstitial component is the main determinant of increased renal vascular resistance. The prevalence of pathological renal RI (≥ 0.70) was significantly higher in patients with isolated TIN (65.2%) than in those with non-IgA TIN (12.3%) and IgAN (32.7%) (P < 0.001). It is interesting to note that IgAN, despite being a primary GN, still has renal RI values intermediate between isolated TIN and non-IgA TIN, probably due to the presence of secondary tubulointerstitial damage frequently associated with this disease[8].

This finding is clinically significant: Approximately two-thirds of patients with pure tubulointerstitial damage already show ultrasound evidence of microvascular impairment at diagnosis. The marked difference compared to the non-IgA-TIN group (only 12.3% with pathological RRI) suggests that glomerular involvement may alter the renal hemodynamic pattern. These data suggest that renal RRI could be evaluated in future studies as a potential screening tool to identify patients with predominant tubulointerstitial damage[25].

Prognostic utility of ultrasound parameters

Ultrasound parameters such as renal echogenicity and cortical thickness offer valuable prognostic insights in glomerular disease. Several studies have investigated the correlation between renal echogenicity and a decrease in renal function. While high parenchymal echogenicity is normal in the neonatal period, in adulthood, higher cortical echogenicity correlates with lower estimated GFR (eGFR) and greater histological damage (fibrosis, sclerosis, atrophy)[20]. Obtaining an objective estimate of renal echogenicity has been challenging before Hricak et al[26] developed an echogenicity grading classification based on the comparison of renal cortex echogenicity to liver and splenic parenchyma, and to renal medulla and renal sinus. Although echogenicity assessment is qualitative and user-dependent and does not help in differentiation of etiological causes of renal damage, this grading found large use, showing correlation with eGFR, serum creatinine, and histopathology, and still represents a helpful non-invasive indicator of pathologic burden and may prove useful in monitoring disease progression during longitudinal evaluation[14,27,28]. A recent study found that the Echogenicity Difference Value between Renal Sinus and Renal Cortex presents better correlation to eGFR decrease than the radiologists’ visual analysis of renal echogenic characteristics[29]. Nonetheless, pitfalls of this technique must be recognized: First of all, a valid assessment involves normal liver echotexture (as liver steatosis can produce false negatives for low-grade disease) while renal intrinsic conditions can alter medullary echogenicity (i.e., nephrocalcinosis, medullary tubular ectasia, or medullary sponge kidney)[20].

A multicenter Italian study (2795 patients, 48 centers) investigated how histopathological diagnosis affects the relationship between kidney ultrasound parameters and eGFR[7]. While bipolar kidney diameter and parenchymal thickness directly correlated with eGFR (R² = 0.064), adding diabetes and proteinuria improved the model (R² = 0.100). Importantly, incorporating histopathological diagnosis substantially increased explanatory power (R² = 0.216), with significant interaction between diagnosis and kidney length (P = 0.006). This demonstrates that the kidney size-eGFR relationship is disease-specific[7]. These findings emphasize that histopathological diagnosis provides crucial information beyond ultrasound measurements alone for comprehensive CKD assessment.

Ultrasound plays a crucial role in differentiating acute from chronic kidney damage and in longitudinal monitoring of disease progression. Chronic GN often presents with small, echogenic kidneys and reduced corticomedullary differentiation. Acute forms, conversely, may show normal or enlarged kidneys with increased echogenicity due to edema or inflammation. Ultrasound can distinguish acute pyelonephritis from lower urinary tract infection based on kidney enlargement. Acute kidney injury may present with enlarged kidneys and preserved parenchymal architecture[30,31]. Serial exams allow tracking of structural parameters such as cortical thickness and echogenicity[32]. Although not sensitive to early histological changes, these indicators help detect progression or flare-ups.

ULTRASOUND FEATURES IN SPECIFIC GN TYPES

While ultrasound provides valuable information, it generally lacks the specificity to definitively diagnose individual GN subtypes, necessitating renal biopsy for confirmation[33].

Lupus nephritis

Conventional ultrasound in lupus nephritis (LN) provides important clinical information by detecting morphological abnormalities and guiding procedures, but its role is limited by the complexity of the disease. LN encompasses six distinct histological classes with markedly different microscopic patterns, yet conventional ultrasound cannot reliably distinguish between them. Grayscale imaging findings such as increased cortical echogenicity, altered kidney size, or loss of corticomedullary differentiation reflect chronic parenchymal damage but lack specificity for histological classification, which always requires renal biopsy[34]. Real-time ultrasound guidance facilitates safer renal biopsies in LN patients, particularly those with thrombocytopenia or requiring anticoagulation[7,35].

Recent technological advances have opened further possibilities. Mediating the use of shear wave elastography (SWE), commonly employed for the indirect assessment of liver fibrosis, Yuan et al[35] employed ultrasound viscoelastic imaging for differentiation of proliferative LN (PLN) from non-PLN (nPLN): Parameters of average elasticity (Emean), tissue viscosity (Vmean), and dispersion coefficient (Dmean) showed significant correlations with both the National Institutes of Health activity index and chronicity index, offering valuable diagnostic and prognostic information[35]. Emean, Vmean, and Dmean were significantly higher in patients with nephritis compared to the control group, while Vmean and Dmean of PLN patients were higher than in the nPLN group (Vmean: 2.27 ± 0.31 Pa•s vs 2.02 ± 0.21 Pa•s; P < 0.05), correlating with histologic activity indices (r = 0.57) and chronicity indices (r = 0.34)[35]. A Vmean cutoff of > 2.16 Pa•s predicted PLN with 86.8% specificity [area under the curve (AUC) = 0.77], improving to AUC = 0.83 when combined with serum creatinine and anti-dsDNA levels[35].

Wei et al[36] conducted a retrospective study exploring how quantitative parameters of CEUS can help distinguish proliferative from nPLN. In 58 biopsy-confirmed patients (38 PLN, 20 nPLN), CEUS performed within three days prior to biopsy showed that patients with PLN had significantly higher absolute time to peak (∆TTP), half-life in the descending phase (DT/2), and area under the time-intensity curve compared to non-proliferative forms (P < 0.05). Area under the time-intensity curve, with a cutoff of 8049.0, achieved an AUC of 0.810, sensitivity of 68.8%, and specificity of 84.6% for identifying proliferative forms. These quantitative CEUS parameters offer non-invasive diagnostic information that complements traditional laboratory markers, with particular value for monitoring and for patients in whom biopsy is contraindicated[36].

Recent advances in ultrasound radiomics and machine learning have enabled potential non-invasive assessment of LN activity. A study by Qin et al[10] demonstrated that a multilayer perceptron model analyzing 10 radiomic features could differentiate active (Activity index > 7) from inactive LN with an AUC of 0.82, sensitivity of 78.9%, and specificity of 69.2%. Key discriminative features included texture patterns reflecting microstructural changes (e.g., fibrosis, inflammation) invisible to conventional B-mode imaging. This approach may provide additional information for monitoring disease progression[10].

IgAN

IgAN represents the most prevalent form of primary GN worldwide[37]. Despite variable clinical presentations, IF, tubular atrophy, and glomerulosclerosis are the most reproducible predictors of progressive renal function decline in IgAN[38]. In IgAN, kidney size seems to inversely correlate with fibrosis severity: A large study of 725 IgAN patients demonstrated renal length was significantly associated with tubular atrophy/IF (TA/IF) in univariate analysis (P < 0.001). In multivariate analysis, renal length was an independent predictor of TA/IF [P < 0.001, odds ratio = 0.927, 95% confidence interval (CI): 0.905-0.95]. Patients without significant fibrosis (T0) had a mean renal length of 104.9 ± 8.4 mm, while those with fibrosis (T1 and T2) had significantly shorter kidneys at 100.5 ± 7.6 mm[39]. However, this remains an indirect and non-specific indicator requiring biopsy confirmation. Renal length negatively correlates with fibrosis and tubular atrophy in IgAN, but on its own has limited diagnostic utility. Turgutalp et al[40] in their prospective study investigated SWE as a non-invasive tool to assess IF and IFTA in IgAN patients. The study included 30 biopsy-proven IgAN patients (group 1) and 28 healthy controls (group 2). The investigators using SWE measured Young’s elastic modulus (YM), a physical measure of tissue stiffness. Significant differences were found between groups in serum creatinine, YM, 24-hour proteinuria, parenchymal thickness, and eGFR (P < 0.05). YM cutoff values correlated with disease severity: 0-15 kPa (absent IF/mild IFTA), 16-27 kPa (moderate IF/IFTA), and > 28 kPa (severe IF/IFTA). The study concluded that SWE-measured YM effectively discriminates between different stages of IF and IFTA in IgAN patients, offering a promising non-invasive monitoring tool[40].

Zhang et al[41] evaluated ultrasound radiomics for differentiating IgAN from membranous nephropathy (MN) using 623 images from 68 patients (46 MN, 22 IgAN). From 180 extracted radiomics features, the Least Absolute Shrinkage and Selection Operator (LASSO) regression selected 33 discriminative features, including gray-level variance, run-length non-uniformity, and wavelet-derived texture parameters. Four machine learning classifiers showed promising results. Random forest achieved the highest AUC (0.7639) and specificity (87.50%), while logistic regression attained the highest accuracy (76.47%) and sensitivity (88.89%). All models exceeded AUC: 0.70. Integrating multiple ultrasound slices per patient improved accuracy by at least 5% vs single-slice analysis, and radiomics significantly outperformed clinical parameter-based models from prior studies[41].

Yang et al[42] in their study evaluated whether CEUS can noninvasively assess TA/IF in IgAN. Eighty biopsy-proven IgAN patients and 33 healthy controls underwent CEUS to quantify renal cortical and medullary perfusion using time-intensity curve parameters. While conventional ultrasound and medullary perfusion showed no significant differences, cortical perfusion metrics, especially peak intensity, AUC, and wash-in slope, were significantly reduced in IgAN and progressively decreased from T0 to T2 according to the Oxford TA/IF classification. Cortical peak intensity independently correlated with TA/IF severity, and receiver-operating characteristic analysis demonstrated strong discriminatory ability, particularly for identifying advanced fibrosis (AUC approximately 0.95)[42].

Similarly, Zhang et al[43] investigated a total of 25 children with IgAN or Henoch-Schönlein purpura nephritis using CEUS. The authors reported that the parameters rise time (RT), TTP, and mean transit time appeared to be related to histological severity. In particular, RT and TTP were higher in children with Lee grade IV than in those with lower grades[43]. Furthermore, regarding the MEST-C classification, RT and TTP values gradually increased with the C grade (presence of crescents) and were markedly higher in the C2 group (crescents in ≥ 25% of glomeruli). No significant differences in CEUS parameters among the different M (Mesangial hypercellularity), E (Endocapillary hypercellularity), S (Segmental sclerosis), or T (Tubular atrophy/interstitial fibrosis) score groups were found[43].

These findings indicate that CEUS-derived cortical perfusion measurements may serve as promising, non-invasive imaging biomarkers of structural damage in IgAN, potentially aiding risk stratification and follow-up alongside traditional clinical and histological tools[41-43].

Interestingly, a large multicenter ultrasomics-based machine-learning model has been developed to non-invasively predict the extent of IF/TA in IgAN. In a cohort of 471 biopsy-proven cases from four institutions, ultrasomics features combined with clinical and conventional ultrasound parameters achieved excellent discrimination between mild and moderate-to-severe IF/TA, with AUC values up to 0.913 in the training cohort and 0.904 in external validation. The integrated Rad-Clinic model outperformed both radiomics-only and clinical-only approaches, and SHapley Additive exPlanations analysis confirmed the biological plausibility of the most informative features. This comprehensive model, therefore, represents a promising non-invasive tool for staging fibrotic burden in IgAN[44]. While promising, these machine learning -based approaches require prospective validation and are not yet ready for routine clinical implementation. External validation across diverse populations and ultrasound equipment remains essential.

From a methodological perspective, most radiomics and artificial intelligence models in GN are based on retrospective single-center datasets, with validation mainly relying on internal strategies such as cross-validation or random train-test splitting. External validation has been reported only in a limited number of studies. Critical issues for clinical applicability include the stability of radiomic features across different acquisition settings, the impact of operator-dependent segmentation on feature extraction, and the performance of trained models when applied to populations with different disease prevalence and histological spectra. Therefore, prospective multicenter studies with predefined imaging protocols and independent validation cohorts are required before these tools can be considered for routine clinical use.

MN

In MN, which typically presents with nephrotic syndrome, ultrasound usually shows normal-sized kidneys with preserved corticomedullary differentiation unless complications develop[45]. There is no specific sonographic pattern for MN, but grayscale ultrasound and colour Doppler can help detect complications such as renal vein thrombosis, a recognized risk in nephrotic states due to urinary loss of anticoagulant proteins. Absent or diminished venous flow signals suggest thrombosis and require confirmatory imaging[46,47]. The RI in uncomplicated MN typically remains normal or only slightly elevated[48].

Serial ultrasound examinations can provide valuable longitudinal monitoring, tracking renal size and cortical thickness to detect progression to CKD in patients with incomplete remission[49]. Real-time guidance also improves biopsy safety in these prothrombotic patients, ensuring adequate sampling and minimising complications[50].

Lu et al[51] in their prospective study, used MicroFlow Imaging (MFI), a novel ultrasound technique, to detect renal microvascular damage in MN. The Vascular Index (VI) quantifies fractional microvascular volume within the renal cortex using Doppler pixel density analysis. In MN, subepithelial immune complex deposition leads to glomerular basement membrane thickening and podocyte effacement. As the disease progresses to CKD, secondary arteriolosclerosis and peritubular capillary rarefaction may develop, further reducing cortical perfusion[51].

Seventy-six participants (38 MN patients and 38 healthy controls) were examined, presenting a VI significantly lower in MN patients compared to controls (0.35 ± 0.18 vs 0.65 ± 0.09, P < 0.001), reflecting the microvascular damage characteristic of the disease. The reported decline in VI represents approximately a 46% reduction in functional capillary density. Importantly, this anatomical rarefaction occurs before compensatory mechanisms are exhausted; hyperfiltration in residual nephrons maintains creatinine-based eGFR until approximately 50%-75% of nephrons are lost. VI may be reduced before functional biomarkers become abnormal, suggesting potential utility in detecting early microvascular changes in CKD. Moreover, MFI demonstrated superior diagnostic performance (AUC: 0.94; 95%CI: 0.89-0.99) compared to serum creatinine (AUC: 0.87; 95%CI: 0.79-0.95) for identifying MN. Patients were stratified by vascular damage severity: Mild (29%), moderate (40%), and severe (31%). In the mild damage group, VI showed better predictive performance (AUC: 0.79; 95%CI: 0.64-0.94) than eGFR (AUC: 0.63; 95%CI: 0.43-0.84). MFI identifies early microvascular alterations when eGFR, the standard parameter for renal function, is still not very sensitive. This suggests potential clinical utility in early diagnosis[51].

Ultrasound elastography, combined with microvascular assessment, shows particular promise in pediatric MN. In a study of 60 children with CKD, 18 of whom had histologically confirmed MN, the YM values of the renal parenchyma were significantly higher in children with stage I and II MN than in healthy controls (P < 0.05)[52]. This increased tissue stiffness reflects the pathological changes characteristic of MN: Thickening of the glomerular basement membrane and immune complex deposition that alter its mechanical properties. Notably, elastography detected these alterations even at stage I, when conventional parameters may still be within normal limits, opening a window for more timely therapeutic intervention. In pediatrics, where limiting biopsy frequency is particularly desirable, this early diagnostic sensitivity may be of clinical value[52].

OTHER PRIMARY AND SECONDARY FORMS OF GN

Different ultrasound techniques can assist in the evaluation of patients with GN by providing valuable functional and structural information, even though they cannot reliably distinguish among specific disease subtypes. In a large Italian biopsy cohort of 2795 patients, ultrasound parameters proved to be closely linked to renal function across the spectrum of GN. The association between ultrasound and function is not uniform because it varies depending on histology[7]. Kidney length and parenchymal thickness were directly correlated with eGFR, with each additional centimeter of bipolar diameter corresponding to an increase of about 6.9-7.8 mL/minute/1.73 m² in eGFR, and each additional millimeter of parenchymal thickness adding approximately 0.7 mL/minute/1.73 m²[7]. However, Ultrasound alone is not sufficient to predict renal function; many glomerulopathies have overlapping ultrasound presentations, and ultrasound variables alone explained only 6.4% of eGFR variability. Histology significantly improves predictive ability because it captures the nature and chronicity of kidney damage: When the histopathological diagnosis was introduced into the model, the explained variance rose sharply to 21.6%. Conditions such as minimal change disease, hereditary glomerulopathies, LN, Henoch-Schönlein purpura nephritis, acute post-infectious GN, MN, and IgAn tended to cluster toward higher eGFR values. In contrast, diseases such as antineutrophil cytoplasmic antibodies (ANCA)-associated vasculitides, membranoproliferative GN, cryoglobulinemic GN, thrombotic microangiopathies, and paraprotein-related disorders were associated with the lowest eGFR[7].

The same renal diameter may indicate different levels of functionality in different pathologies. A significant interaction emerged between diagnosis and kidney length (P = 0.006), demonstrating that the structural–functional relationship observed on ultrasound depends critically on the underlying disease. Notably, some of these entities may still present with normal or even enlarged kidneys despite low eGFR[7].

In the study of Nestola et al[53], the authors evaluated 31 patients with early chronic GN and preserved renal function; they used CEUS to derive time-intensity curves of renal perfusion and compared them with histological parameters of disease activity. They found that persistence of contrast signal during the wash-out phase of CEUS correlated significantly with overall histologic activity (P = 0.016) and particularly with mesangial hyperplasia (P = 0.008). This suggests that CEUS is able to detect microvascular alterations associated with active inflammation but does not establish a causal relationship and may reflect hemodynamic changes common to various GNs[53]. No significant correlations were observed between time-intensity-curve parameters and clinical or Doppler ultrasound data. This result is expected: With normal renal function, filtration indices do not yet reflect early histological damage. This makes CEUS potentially useful in the pre-clinical phase of the disease; a prolonged wash-out phase may reflect glomerular capillary perfusion disturbance in early GN, and CEUS may serve as a non-invasive imaging biomarker of glomerular disease activity[53].

In the prospective study of Galesić et al[54], 50 patients with biopsy-proven glomerular diseases and 60 healthy controls underwent duplex Doppler sonography to measure intrarenal arterial RI. The mean RI in the glomerular disease group was 0.68 ± 0.09, significantly higher than in controls (0.596 ± 0.035; P < 0.01). The difference is significant and indicates an increase in intrarenal vascular resistance in patients with glomerular diseases. However, the intervals overlap: This reduces the RI’s capacity as an individual diagnostic test, while showing a pathological trend at the group level. Patients with Membranoproliferative GN had the highest mean RI (0.817 ± 0.062), and this difference across GN subtypes was statistically significant (P = 0.022). Such a high average value confirms marked microvascular impairment in this form. Despite statistical significance, the sample size for each subtype is presumably small, so generalizability is limited. IR is more useful as a marker of severity or chronicity than as a diagnostic differentiator. Its low specificity limits its use in distinguishing between subtypes of GN[54].

Sugiura et al[55] expanded this concept by evaluating not only RI but also a morphometric “atrophic index” (renal sinus length/renal length), directly comparing these imaging markers with histological tubulointerstitial injury. In their prospective study involving 60 patients with primary or secondary glomerular diseases, Doppler ultrasonography was performed just before renal biopsy to assess whether intrarenal hemodynamic indices could detect tubulointerstitial damage.

Introducing a morphometric index, such as that of the renal sinus, may increase sensitivity in identifying anatomical changes associated with fibrosis. Receiver-operating characteristic analysis found an RI cut-off of 0.65 discriminated tubulointerstitial changes with 100% specificity and 57.1% sensitivity, while an atrophic index cut-off of 0.70 gave specificity 100% and sensitivity 61.9%. When both indices were combined (either or both elevated), sensitivity increased to 85.7%, and specificity remained high (94.4%). This is the most promising finding: The combination significantly improves the ability to identify patients with histological tubulo-interstitial lesions, while maintaining very high specificity. This suggests that the integration of multiple ultrasound parameters (one hemodynamic, one morphometric) can overcome the limitations of individual indices and offer greater reliability in identifying damage[54,55].

Overall, these data show that ultrasound is an essential imaging tool that provides valuable information when interpreted alongside histological findings. It can quantify renal perfusion, vascular resistance, and parenchymal stiffness, evaluating subtle changes that correlate with disease activity or chronicity. However, ultrasound markers are not specific enough to distinguish between different subtypes of GN or to define the underlying immunopathological mechanisms. Histology, by contrast, provides a definitive assessment of glomerular, vascular, and tubulointerstitial lesions. When the two approaches are integrated, ultrasound offers a non-invasive, repeatable method to monitor changes over time, while biopsy remains the principal diagnostic and prognostic tool within which these imaging findings can be correctly interpreted. The use of both methods optimises diagnostic accuracy, improves clinical decision-making, and supports more personalized management of glomerular diseases (Tables 1, 2, and 3).

Table 1 Prognostic significance of key ultrasound morphologic parameters in glomerulonephritis.

Parameter1	Typical US finding	Correlation with histology	Predictive value2
Renal echogenicity	Increased cortical echogenicity (grades II-IV)	Glomerulosclerosis, tubulointerstitial fibrosis, inflammation	Higher echogenicity correlates with lower eGFR
Cortical thickness	Decreased cortical thickness	Positive correlation with eGFR	< 4.0 mm/cm predicts > 30% eGFR decline or dialysis initiation (sensitivity 72.5%, specificity 80%)

¹Renal cortical thickness normalized for patient height: Cortical thickness (mm) divided by height (cm).

²Reported cut-off values are protocol dependent and may vary according to acquisition settings and patient population.

eGFR: Estimated glomerular filtration rate; US: Ultrasound.

Table 2 Practical ultrasound features in differentiating acute vs chronic glomerulonephritis¹.

Parameter	Acute GN/AKI	Chronic GN/CKD	Key differentiating feature
Kidney size	Normal or enlarged	Decreased renal length (< 8 cm)	Enlarged = acute; small = chronic
Cortical thickness	May be normal or increased	Reduced	Thinning indicates chronic damage
Echogenicity	Increased (infiltrates/edema)	Increased (fibrosis/sclerosis)	Alone non-specific; paired with size aids differential diagnosis
Corticomedullary differentiation	May be increased, Preserved or increased	Often reduced or lost	Loss of differentiation implies chronicity

¹Reported thresholds are protocol dependent and may vary according to acquisition settings and patient population.

AKI: Acute kidney injury; CKD: Chronic kidney disease; GN: Glomerulonephritis.

Table 3 Ultrasound techniques in glomerulonephritis¹.

Ultrasound technique	Technical principles	Primary applications in GN	Key findings/parameters	Ref.
B-mode (grayscale) ultrasound	Reflection of ultrasound waves from tissue interfaces generates 2D anatomical images	Morphological assessment: Kidney size, cortical thickness, echogenicity, corticomedullary differentiation	Normal kidney: 10-12 cm length, cortical thickness 7-10 mm; increased echogenicity correlates with fibrosis, glomerulosclerosis; cortical thickness < 4.0 mm/cm predicts eGFR decline	O’Neill et al[22], 2000; Moghazi et al[14], 2005; Petrucci et al[6], 2018; Andrulli et al[7], 2024
Color/power doppler ultrasound	Detection of blood flow via Doppler shift; power Doppler measures flow magnitude independent of direction	Vascular resistance assessment, large vessel abnormalities detection	RRI = (PSV - EDV)/PSV; RRI ≥ 0.70 indicates tubulointerstitial damage; isolated TIN: RRI 0.73 vs TIN + GN: 0.64; 65% of isolated TIN shows pathological RRI	Yura et al[24], 1993; Gigante et al[8], 2016; Gigante et al[25], 2022; Galesić et al[54], 2004
Contrast-enhanced ultrasound (CEUS)	Microbubble contrast agents (sulfur hexafluoride) enhance vascular visualization; time-intensity curves quantify perfusion	Perfusion assessment, disease activity monitoring, microvascular alterations detection	Prolonged wash-out correlates with histological activity (P = 0.016); mesangial hyperplasia correlation (P = 0.008); TIC-AUC cutoff 8049.0 - arbitrary units for PLN vs nPLN (AUC 0.810)	Nestola et al[53], 2018; Wei et al[36], 2025; Yang et al[42], 2020; Qasim et al[56], 2025
Shear wave elastography (SWE)	Acoustic radiation force generates shear waves; tissue stiffness measured as Young’s modulus (YM) or shear wave velocity (SWV)	Fibrosis detection and quantification, chronic changes assessment	YM values in CKD significantly higher than controls; YM cutoffs: 0-15 kPa (absent/mild IFTA), 16-27 kPa (moderate), > 28 kPa (severe); cutoff < 20.77 kPa for fibrosis detection; negative correlation with eGFR (r= -0.576, P < 0.0001)	Turgutalp et al[40], 2020; Huynh et al[57], 2022; Choi et al[58], 2023; Sofia et al[59], 2017; Grenier et al[60], 2011
Viscoelastic ultrasound imaging	Measures both elastic (storage) and viscous (loss) properties of tissue using plane-wave ultrasound	Differentiation of proliferative vs non-proliferative lupus nephritis, tissue characterization	Vmean and Dmean elevated in proliferative LN; cut-off vmean 2.16 Pa·s: AUC 0,77, sensibility 56.7%, specificity 86.8% per PLN; combined model: (Vmean + Scr + anti-dsDNA) AUC 0.83	Yuan et al[35], 2025
Microvascular flow imaging (MVFI)	Advanced Doppler techniques (SMI, MFI, MicroFlow) using clutter suppression and adaptive filtering to detect slow-flow microvessels	Microvascular perfusion quantification, early vascular damage detection	Vascular index (VI) = Doppler pixels/total pixels; VI in MN: 0.35 ± 0.18 vs controls: 0.65 ± 0.09 (P < 0.001); 46% reduction in functional capillary density in MN; higher diagnostic performance than eGFR in early disease, AUC VI (0.79), AUC eGFR (0.63)	Lu et al[51], 2024; Qin et al[61], 2014
Ultrasound radiomics	Extraction of quantitative texture features from ultrasound images using computational algorithms	GN subtype classification, fibrosis prediction, pathological correlation	180 features extracted from renal parenchyma; LASSO regression selects discriminative features; gray-level variance, run-length non-uniformity, wavelet-derived textures most informative	Zhang et al[41], 2021; Qin et al[10], 2023; Floreani et al[62], 2021
Machine learning/deep learning ultrasomics	Neural networks analyze ultrasound images combined with radiomics features and clinical data	Integrated diagnostic models, fibrosis staging, outcome prediction	U-net for automatic segmentation; random forest (RF) for classification (37 features); combined nomogram models with clinical factors; SHAP analysis for feature interpretation	Vernuccio et al[63], 2020; Huang et al[44], 2025; Kawashima et al[64], 1997; Stunell et al[65], 2007

¹The table summarizes key ultrasound modalities used in glomerulonephritis, including B-mode, Doppler, contrast-enhanced ultrasound, elastography, viscoelastic imaging, MicroFlow imaging, and radiomics, highlighting their main quantitative parameters, diagnostic thresholds, and correlations with renal pathology and function.

2D: Two-dimensional; eGFR: Estimated glomerular filtration rate; RRI: Renal resistive index; PSV: Peak systolic velocity; EDV: End diastolic velocity; TIN: Tubulointerstitial nephritis; GN: Glomerulonephritis; CEUS: Contrast-enhanced ultrasound; SWE: Shear wave elastography; YM: Young’s modulus; SWV: Shear wave velocity; CKD: Chronic kidney disease; LN: Lupus nephritis; AUC: Area under the curve; PLN: Proliferative lupus nephritis; MVFI: Microvascular flow imaging; SMI: Superb microvascular imaging; MFI: MicroFlow imaging; VI: Vascular index; LASSO: Least absolute shrinkage and selection operator; SHAP: SHapley Additive exPlanations; RF: Random forest.

ROLE OF CT AND MRI IN GN DIAGNOSIS

Due to the molecular and immune-mediated pathological mechanisms, the microscopic features of tissue damage, and the diffuse organ involvement, GN lacks pathognomonic imaging features that provide a definitive diagnosis. As a result, even in the era of advanced imaging, renal biopsy remains the gold standard for the diagnosis and classification of GN, so the role of imaging is not emphasized in the most recent guidelines on the management of glomerular diseases. In fact, radiological assessment is currently limited to identifying hemorrhagic and obstructive complications[56].

In fact, for many years, the role of imaging in the management of glomerular diseases has been limited to renal morphometry, detection of complications or collateral extrarenal diagnostic elements, and guidance of diagnostic procedures such as renal biopsy. The ultrasound plays a primary role in percutaneous puncture guidance, followed by computed tomography (CT)[5].

While morphological evaluations such as size and parenchymal thickness are easily obtained by ultrasound, panoramic imaging methods like CT and MRI, especially with contrast media (CM), provide operator-independent, detailed descriptions of parenchymal features such as inflammatory changes, impaired perfusion, and extrarenal complications, including renal vein thrombosis, hemorrhage, and abscesses. Their comprehensive imaging capabilities also support differential diagnosis, including the identification of neoplastic causes of obstruction.

CT is an imaging technique based on tomographic reconstruction using volumetric analysis of X-ray absorption, primarily relying on photoelectric attenuation of radiation. The resulting images can therefore be described as a mono-parametric technique that allows assessment of tissue density. CT has established its role over ultrasound mainly in acute and emergency settings due to its ability to assess morphology, identify causes and complications, support differential diagnoses, enable visual preoperative planning in oncology, and assess organ perfusion with the use of CM[57-60]. Quickly covering large volumes, CT also provides essential information on distant pathological features, facilitating diagnosis in systemic diseases such as immunoglobulin G4 (IgG4)-related and lupus-related disease[61,62].

MRI is a non-invasive imaging modality that provides excellent soft tissue contrast and does not use ionising radiation. It is particularly valuable for serial follow-up or when iodinated CMs are contraindicated. MR images are produced from the echo signal of hydrogen nuclei in organic molecules, enabling multiparametric tissue assessment, including information on inflammation, fibrosis, and perfusion.

Given the lack of specificity in imaging-based diagnosis of GN, it is essential to clarify the potential roles of imaging in the evaluation of glomerular diseases. Although not sufficient for etiological diagnosis, imaging remains vital for assessing the extent of kidney involvement, monitoring treatment response, and supporting differential diagnosis with high accuracy. Current literature does not describe GN-specific inflammatory imaging features on CT or MRI; instead, it focuses on general inflammatory injury to the kidney, mainly secondary to other causes, with extensive descriptive literature on pyelonephritis and infectious diseases. Accurate interpretation of pathological imaging features is not possible without proper correlation with symptoms and laboratory testing, due to the overlap of pathological findings among different etiologies.

CT: Morphologic assessment

Signs of inflammatory involvement of the kidneys are variable on imaging and depend on the extent of disease, which may be localised or diffuse. In acute diffuse parenchymal injury, unenhanced CT may reveal suggestive signs such as renal enlargement, stranding of perinephric fat due to edema or fluid, and thickening of Gerota’s fascia[63]. Another subtle sign of renal inflammation is blurring of the renal sinus fat, secondary to compression and fluid thickening[64]. The use of CM is necessary for the assessment of vascularization and parenchymal enhancement. Inflammatory infiltration leads to edema, compression, and ischaemia, resulting in pathological enhancement[64]. In addition to parenchymal swelling, visible even on the unenhanced scan, diffuse renal injury is characterized by decreased or delayed enhancement with loss of cortico-medullary differentiation; involvement of extrarenal tissues with fat stranding is common[63].

In ascending infections such as pyelonephritis, the nephrographic phase shows areas of hypoenhancement with a patchy distribution[63]. These areas typically assume a triangular shape, extending from the papilla to the cortex. Consistent parenchymal swelling may also present as a pseudonodular tumefactive area, requiring differential diagnosis from tumoral tumefaction[65]. On a delayed scan, the abnormal areas may show a dense nephrogram, which persists for a long time after the normal parenchyma has washed out the enhancement. This radiological sign is commonly called a striate nephrogram[65]. This sign is also common to other conditions that require accurate differential diagnosis; the most common causes may include ischaemia, contusion, tubular obstruction (for example, in cases of myoglobinuria), and urinary obstruction[66]. Differential diagnosis may be guided by additional features: Coexistent thickening of the collecting system indicates pyelitis; the presence of the peripheral rim sign is distinctive for renal infarction; pericapsular fluid suggests inflammatory infiltration, while a hematoma suggests a traumatic etiology. In diffuse inflammatory damage, persistence of the nephrogram during the delayed phase, even hours after CM administration, is also possible. A persistent nephrogram has been documented in cases of interstitial nephritis and other conditions such as bilateral obstructive uropathy, renal artery or renal vein stenosis, systemic hypotension, and acute tubular necrosis[67]. It is crucial to distinguish striate nephrograms from conditions with similar imaging appearances, such as medullary sponge kidney (Lenarduzzi-Cacchi-Ricci disease), which shows the characteristic “bouquet sign” on excretory urography, due to the paintbrush appearance of dilated medullary tubuli during excretory urography[68].

Separate mention must be made for IgG4-related kidney disease (IgG4-KD), due to the specific patterns of renal damage described for this systemic immune disease[69]. IgG4-KD patterns have been classified as parenchymal, pelvic, and perinephric, with renal localisation being the most common and often accompanied by coexisting localizations in the same individual. Imaging patterns of renal localization of IgG4-KD include solitary nodular lesions, multiple nodules, and diffuse patchy distribution[69].

Parenchymal nodules typically appear hypoattenuating on the corticomedullary phase, with progressive enhancement during the delayed phases. Their shape ranges from nodular to wedge-shaped. Solitary lesions often require differential diagnosis with primary and secondary neoplastic nodules, which often require histopathological proof, while multiple lesions, particularly when wedge-shaped, may be mistaken for pyelonephritic localizations[70]. Perinephric localization of IgG4-KD usually appears as a hypoattenuating tissue band surrounding the kidneys[69]. This rare disease requires careful differential diagnosis from renal lymphoma; therefore, features such as increased serum IgG4 Levels and response to steroid treatment become essential[71]. Chronic progression of inflammatory renal injury leads to fibrotic scarring and permanent loss of function. Focal pyelonephritis may cause cortical retraction due to reduced parenchymal thickness with hypoattenuating scar tissue. Global atrophy of the kidney with compensatory contralateral kidney hypertrophy occurs in diffuse disease[72].

MRI morphologic assessment

When available, MRI can demonstrate the same findings as CT in inflammatory renal disease, such as swelling, perfusion abnormalities, and hypoenhancing regions. In addition, most inflammatory changes in the kidneys can be detected by MRI with significantly higher sensitivity compared to CT and without the use of CM. The high contrast differentiation between fluid and fat is a strategic feature, allowing selective visualization of fluid infiltrate on T2 sequences. Fat suppression techniques further increase MRI sensitivity to acute inflammatory changes by maximising the signal of fluid content on unenhanced scans.

The use of DWI reveals restricted diffusion of extracellular free water due to edema with high sensitivity. The diagnostic power of DWI in detecting inflammatory renal changes has been widely confirmed in both infective and immune-mediated diseases[73,74]. DWI provides a powerful tool for the diagnostic management of patients who cannot receive CM injection, enabling diagnosis without ionising radiation or contrast agents. When required, a T1 Gadolinium chelate CM study completes the MR examination with the same enhancement kinetics described for CT, depicting round or wedge-shaped hypoattenuating areas corresponding to the hyperintense regions seen on T2 imaging[69].

Gadolinium-enhanced MRI is also valuable for vascular mapping, which is crucial in kidney donor evaluation, especially for identifying accessory vessels or abnormal anatomy, such as extrahilar branching or retrocaval vessel position[75]. For this specific use, non-contrast techniques such as time-of-flight angiography may be sufficient but can present potential pitfalls in cases of low-calibre accessory branches, whereas CM angiography with fast gradient-echo imaging allows precise detection of vascular anomalies with good temporal resolution[76].

MRI functional assessment

Morphological assessment represents only the tip of the iceberg when considering the potential of MRI imaging. Advanced functional imaging techniques (fMRI) can track changes in kidney structure and function with the reproducibility necessary for serial assessment over time to monitor treatment response. Several techniques combine structural mapping and functional assessment, each with different applications.

DWI assesses the free Brownian motion of extracellular water within tissues. Microstructural barriers, such as cellular membranes, may restrict the free diffusion of extracellular water, influencing the speed of diffusion. Measurements with different gradient durations and amplitudes (P-values) allow the creation of a diffusion map, measuring the apparent diffusion coefficient (ADC), which quantifies the speed of water diffusion in tissues. In addition, diffusion tensor imaging can assess the directionality of this movement, providing information about tissue anisotropy, a parameter strongly influenced by structural arrangement[77]. For example, diffusion of water is influenced by the linearity of tubular structures in the medullary part of the kidney, compared with the less ordered structure of the cortex[76]. Processes like renal fibrosis involve both impaired perfusion and water diffusion and scar destruction of ordered structures of the renal parenchyma, making DWI suitable for indirect fibrosis assessment[78].

Many studies have compared DWI/ADC measurements with renal biopsy and renal function in patients with CKD: Cohort studies by Li et al[79] and Zhao et al[80] found a negative correlation between ADC values and renal biopsy results in patients with chronic GN. Another study conducted on a cohort of kidney transplant patients demonstrated lower ADC values in both the cortical and medullary regions of the kidney[79,80], as well as reduced fractional anisotropy values in the medulla, with an inverse correlation with fibrosis observed at biopsy[81]; these imaging parameters, however, did not correlate with other histological changes. These results support the use of DWI/diffusion tensor imaging as a tool to assess fibrosis, which restricts water diffusion, but also demonstrate that this technique cannot differentiate between different pathological mechanisms.

A recent study by Zhang et al[82] involving 97 patients with IgAN, focal segmental glomerulosclerosis, and diabetic kidney disease provided interesting results on the correlation between DWI standard ADC values, eGFR, and the histological degree of fibrosis, assessed through peritubular capillary density and extracellular matrix expansion[82]. The results of this study suggest that lower eGFR and DWI-fraction values are correlated with reduced peritubular capillary density, which is also associated with a higher grade of fibrosis in the kidney. This study also found a negative correlation between extracellular matrix volume and MR elastography values. Similar data on DWI have been reported by Villa et al[83] in a multiparametric MRI study comparing normal patients with CKD patients.

Finally, a recent prospective study on membranoproliferative GN on 7 patients compared fMRI to renal biopsy and clinical data over a time of 1 year: Interestingly comparation of MRI data did not show a significant correlation with fibrosis histologic assessment[11]. Despite several promising results, the relative availability of DWI sequences, its non-invasive nature, and repeatability, several aspects limit the extensive use of DWI in the assessment of renal fibrosis. First, longitudinally validated data with greater value than laboratory testing are still not available for clinical use[84]. In addition, ADC measurement depends on changes in the fluid content of renal tissue, being influenced by hormonal feedback, blood flow, and diuretic or antihypertensive medications, which affect glomerular filtration and hemodynamic status[78].

With arterial spin labeling (ASL) imaging, radiofrequency pulsed gradients are exploited to selectively tag inflowing blood entering the volume slab, allowing to estimate tissue flow information without the use of CM through a preliminary mask subtraction[85]. The main advantage of the ASL technique is its repeatability without the use of CM, making it the ideal method for assessing parenchymal perfusion over time. A comparative study investigated ASL-derived microvascular flow rates in patients with CKD, comparing the results with those from healthy volunteers, and demonstrated a significantly decreased blood flow in the CKD patients[86]. A recent study by Lu et al[86] confirmed a significant difference in renal perfusion with 3T-MRI ASL both in the cortex and in the medulla in 48 CKD patients compared to healthy volunteers. Heusch et al[87] studied a population of 98 renal transplant patients demonstrating a decreased renal perfusion in subjects with CKD and reduced eGFR. Ren et al[88] also assessed renal allograft function early after transplantation using both intravoxel incoherent motion and ASL MRI, demonstrating a positive correlation between eGFR, ADC values, and cortical blood flow, with lower values of ADC and perfusion in patients with impaired graft function. Literature reports show sufficient data to consider ASL MRI a reliable way to estimate renal perfusion in a non-invasive manner. Cortical capillary depletion is associated with fibrotic changes; therefore, ASL MRI could also surrogate fibrosis assessment techniques[88]. However, no studies on the human population with histologic comparison have been conducted in this field, and the only studies in this direction have evaluated flow modification after ischemic injury in murine tests[89].

Technical limitations continue to restrict the widespread use of ASL in routine renal vascularization assessment. The main challenges are the reduced signal-to-noise ratio achievable with the common 1.5T scanners and lengthy acquisition times. Further loss of resolution results from decreased cortical thickness in CKD patients, while the relatively lower percentage of medullary flow compared to cortical flow, even in healthy individuals, and difficulties associated with respiratory triggering, also reduce image quality and resolution. The development of technical solutions offering higher resolution and shorter acquisition times will support the adoption of this technique in daily practice. Investigation of the potential of ASL in fibrosis assessment currently represents a promising research area[90]. AI integration on new platforms will certainly help in this field, enabling higher image quality with reduced acquisition time.

Another MRI technique mediated from brain imaging, BOLD MRI, allows for assessment of renal oxygenation. This technique does not require the use of Gadolinium CM, founding its principles on the paramagnetic properties of deoxygenated hemoglobin, which determines a reduction of the effective transverse relaxation time (T2*), due to the magnetic susceptibility effect on tissues[78]. BOLD MRI imaging is able to detect changes in the oxidation state of hemoglobin in tissues in case of parenchymal hypoxia. The first application to renal oxygenation has confirmed differences in metabolism of renal medulla, which works in a near-hypoxic condition and therefore shows a lower T2* signal compared to the cortex[91]. Despite several studies hypothesizing the correlation of renal hypoxia with CKD, the emerging literature demonstrates that this correlation is not linear, and significant rates of hypoxia increase only in end-stage renal disease[92]. A possible explanation is that a decrease in glomerular filtration due to glomerulosclerosis also involves a reduced oxygen uptake, normalizing results of BOLD oxygenation[78].

On the other side, this technique has been proven to be sensitive to oxygenation changes determined by the use of diuretic drugs and non-steroidal anti-inflammatory drugs. Furosemide, in fact, blocks the active sodium reabsorption, then reducing oxygen consumption and consequently improving medullary oxygenation[78]. Several studies have demonstrated that this drug-induced effect fails in cases of renal fibrosis and arterial hypertension, suggesting the potential of BOLD imaging combined with furosemide to assess renal injury[93].

BOLD alone is probably not able to provide a valuable assay of renal function, but its use alongside other MRI techniques and functional biomarkers still holds great potential. This has been emphasised by recent society statements recognising the need to standardise MRI techniques for renal function evaluation, with an encouraging outlook for the near future[13,94].

Phase-contrast MRI is a non-CM MRI technique that allows the determination of blood velocity and flow in a specific vessel, allowing a reliable quantification separately for each kidney. Studies have demonstrated reliability and good correlation of phase-contrast MRI with other standard methods for renal blood flow measurement, both in vitro and in vivo[95]; the great variability of results due to different technical acquisition protocols has brought to the preparation of position statements to standardize this technique in the assessment of renal flow[96]; this point represents a cornerstone in the field of renal MRI pushing for optimization and standardization of the technique.

Other MRI techniques under investigation, which remain at a speculative level of evidence in human studies, are T1 and T2 mapping. These respectively measure changes in longitudinal and transverse relaxation times. Increases in T1 relaxation have been correlated with fibrosis, as well as edema and cellular swelling, while tissue T2 mapping increases in response to inflammation and tends to decrease with severe fibrosis[97].

The complex assessment of renal injury in GN remains closely linked to histological evaluation, but additional benefits may arise from non-invasive MRI techniques, which also allow repeated measurements to monitor disease progression and treatment response over time. Researchers should not seek a single perfect technique to assess renal function, but rather rely on a multiparametric assessment integrated with all clinical and laboratory data[13].

International societies are making significant efforts to raise awareness and standardize multiparametric MRI protocols in large sample populations, such as the AFIRM (Application of Functional Renal MRI to Improve Assessment of Chronic Kidney Disease) study, which is collecting data on 450 patients[98]. Future efforts will be required not only for standardization but also for the integration of automated measurement and reconstruction tools that combine artificial intelligence with fMRI.

NUCLEAR MEDICINE IN GN

Although nuclear medicine does not play a primary diagnostic role in GN, renal scintigraphy provides functional information, such as clearance, differential function, and perfusion, which are complementary in specific forms of GN. In particular, scintigraphy with dimercaptosuccinic acid labeled with Technetium-99m (^99mTc-DMSA) is a reference method for evaluating the renal parenchyma: Approximately two hours after injection, DMSA shows a renal uptake of 40%-65%, corresponding to the fraction of the radiopharmaceutical retained by the functioning parenchyma[99]. This high and prolonged uptake, supported by an effective half-life of approximately 6 hours in the kidneys, allows for excellent anatomical definition of the cortex, making DMSA particularly suitable for identifying parenchymal lesions and evaluating the differential function of the kidneys. The radiopharmaceutical binds with high affinity to proximal tubular cells, thus explaining its specificity for visualizing the cortical parenchyma[100].

Comparative studies have reported high correlation between ^99mTc-DMSA and dynamic renal scintigraphy with ⁹⁹Tc-mercaptoacetyltriglycine (MAG3), showing complete concordance in 85% of patients and correlation coefficients of r = 0.981 (P < 0.001) for ⁹⁹Tc-labeled diethylenetriaminepentaacetic acid and r = 0.918 (P < 0.001) for MAG3[101]. MAG3, which exhibits 40% greater plasma clearance than ⁹⁹Tc-labeled diethylenetriaminepentaacetic acid[102], has demonstrated sensitivity of 88%-89% and specificity of 88%-100% in detecting cortical defects compared to DMSA[103]. These two radiopharmaceuticals differ in several important respects: The mechanism of action, since DMSA selectively binds to proximal tubular cells while MAG3 is mainly secreted tubularly; acquisition times, with DMSA requiring delayed images 2-4 hours after injection and MAG3 relying on dynamic acquisitions in the first 20-30 minutes. Moreover, there are conditions such as obstruction or urinary stasis, which can significantly alter the behaviour of MAG3 but not that of DMSA. These differences explain why, despite often being correlated in their results, the two tracers remain complementary in the functional and morphological evaluation of the kidney[104,105]. In any case, although several studies report the potential usefulness of these radiotracers in assessing typical features of kidney damage, extensive focused studies are needed to establish the role of these radiotracers for GN.

While traditional renal scintigraphy mainly provides functional and structural information, PET imaging, particularly ¹⁸F fluoro-2-deoxy-D-glucose (¹⁸F-FDG) positron emission tomography (¹⁸F-FDG-PET/CT), offers a window into systemic and renal inflammatory activity, especially in vasculitis-associated GN. FDG-PET/CT can be a valid tool for monitoring systemic disease activity: This imaging technique gathers in activated inflammatory cells, such as macrophages and neutrophils, due to their increased glucose transporter expression. According to meta-analyses of FDG-PET/CT in large vessel vasculitis, the pooled sensitivity at 75.9% (95%CI: 68.7-82.1) and the specificity of 93.0% (95%CI: 88.9-96.0) for detecting active disease[106]. Notably, this technique shows high performance in giant cell arteritis (sensitivity 80%, specificity 89%)[107] and Takayasu arteritis (sensitivity 93%, specificity 92%)[108]. In ANCA-associated vasculitides, which frequently involve GN, FDG-PET demonstrated 100% positivity in granulomatosis with polyangiitis compared to 50% in other ANCA-vasculitides (P = 0.05), with median SUVmax of 5.0 in granulomatosis with polyangiitis vs 2.5 in microscopic polyangiitis/eosinophilic granulomatosis with polyangiitis (P = 0.08)[109]. PET identified renal parenchymal involvement in patients with extracapillary GN, though sensitivity may be reduced in isolated microscopic polyangiitis[109]. For treatment monitoring in large vessel vasculitis, FDG-PET/CT demonstrates moderate diagnostic accuracy, showing a pooled sensitivity of 77% (95%CI: 57%-90%) and specificity of 71% (95%CI: 47%-87%)[110]. The PET Vascular Activity Score, a quantitative score that assesses uptake in multiple vascular areas, demonstrates an AUC of 0.73 for differentiating clinically active from inactive disease, with PET Vascular Activity Score ≥ 10 providing 60.8% sensitivity and 80.6% specificity[111]. Even with these encouraging findings, renal biopsy still stands as the gold standard for a definitive diagnosis of GN, while nuclear medicine techniques mainly play a supportive role in assessing function and tracking systemic inflammatory activity in vasculitic GN.

In addition to ¹⁸F-FDG and traditional renal scintigraphy, novel PET radiotracers are stepping up as promising options for non-invasive evaluation of renal parenchymal disease and fibrosis in GN. The ⁶⁸Ga-PSMA-11 PET/CT, which was originally designed for imaging prostate cancer, shows significant uptake in the renal proximal tubular cells and has revealed promising potential for imaging the renal cortex[112]. In the kidney, PSMA is specifically localized in the brush border and apical cytoplasm of proximal tubular cells, even it is not expressed in the glomeruli or Bowman’s capsule. This localization has been demonstrated by immunohistochemical studies and explains the intense renal uptake seen on PSMA-PET scans. Tubular expression of PSMA is functional and constant, making this radiotracer interesting for imaging the cortical parenchyma and could be a basis for studies on tubular involvement in certain GN.

In comparative studies with ^99mTc-DMSA, ⁶⁸Ga-PSMA-11 PET provided superior image resolution and reduced acquisition time (60 minutes post-injection vs 180 minutes post-injection), with excellent visualization of renal cortical defects in pyelonephritis[113,114]. Split renal function assessment using ⁶⁸Ga-PSMA-11 PET demonstrated high correlation with ^99mTc-MAG3 scintigraphy (r = 0.91, P < 0.001) in 97 patients, with Bland-Altman analysis revealing excellent agreement (mean difference 0.002, 95%CI: -0.110 to 0.114)[115]. However, molecular cortex volume showed only moderate correlation with eGFR (R² = 0.231, P < 0.001), as PSMA expression predominantly reflects tubular rather than glomerular function[112]. PSMA-PET demonstrates high correlation with MAG3 for the assessment of differential function in patients who already undergo PSMA-PET for other indications.

FAPI PET imaging has demonstrated correlation with fibrosis severity in detecting renal fibrosis, a common sequela of chronic GN. FAP is expressed by activated fibroblasts and myofibroblasts, who mediate extracellular matrix accumulation in renal fibrosis. In this context FAPI binds specifically to FAP, so the uptake reflects fibrotic activity. The ⁶⁸Ga-FAPI-04 PET/CT scans revealed a progressive increase in SUVmax that matched the severity of fibrosis: Mild fibrosis had a value of 3.92 ± 1.50, moderate fibrosis was at 5.98 ± 1.60, and severe fibrosis reached 7.67 ± 2.23 (P < 0.001) across 13 patients with confirmed renal fibrosis[116]. In a multicenter study involving 81 patients, there was a negative correlation between GFR and ⁶⁸Ga-FAPI uptake (both SUVmax and SUVmean). Interestingly, neither ⁶⁸Ga-DOTATOC nor ⁶⁸Ga-PSMA showed any correlation with CKD stage, indicating that FAPI specifically binds to activated fibroblasts rather than just being retained non-specifically[117].

In IgAN, ¹⁸F AlF-NOTA-FAPI-04 PET/CT uptake correlated positively with IF and tubular atrophy scores (r = 0.637, P < 0.05), tubulointerstitial inflammation (r = 0.593, P < 0.05), and immunohistochemical staining for α-smooth muscle actin and FAP (all P < 0.01)[106]. These findings are particularly important because they demonstrate correlation with the Oxford classification, the standardized histological grading system for IgAN, as well as with immunohistochemistry for α-SMA (a myofibroblast marker) and FAP, providing substantial triple validation (morphology, immunohistochemistry, and imaging). Patients with mild IF/TA (score = 1) showed significantly lower SUVmax compared to those with moderate-severe IF/TA (score ≥ 2): 3.6 ± 1.1 vs 5.8 ± 0.4 (P = 0.003). These data allow stratification of IgAN patients by severity of IF/TA, with a potential SUVmax cutoff of approximately 4.5-5.0 distinguishing mild from moderate-severe IF/TA. This is clinically important, as IF/TA is the strongest prognostic predictor in IgAN[12].

In LN, baseline FAPI PET imaging demonstrated superior predictive performance compared to FDG-PET for treatment response: FAPI visual assessment showed 85% accuracy, 90% specificity, and 89% positive predictive value vs FDG’s 70%, 50%, and 64% respectively in 20 patients[118]. FDG uptake reflects increased carbohydrate metabolism associated with acute inflammation and activated immune cells, whereas FAPI uptake reflects fibroblast activation related to fibrosis and chronic tissue remodeling. In the context of LN, fibrosis is a more critical determinant of prognosis than acute inflammation, and FAPI provides prognostic information beyond standard clinical and laboratory parameters. FAPI demonstrates correlation with treatment response and may provide additional prognostic information in conjunction with renal biopsy in predicting response to LN treatment. In fact, biopsy evaluates a static moment, while FAPI can reflect the dynamism of the disease. Higher baseline renal FAPI SUVmax correlated with lower eGFR (r = -0.561, P = 0.010) and higher ESR (r = 0.492, P = 0.028), and complete responders demonstrated significantly lower baseline FAPI SUVmax and target-to-background ratio compared to non-responders (P < 0.01)[118]. In multivariate analysis, negative FAPI uptake emerged as the sole independent predictor of complete response to induction therapy (P < 0.01)[118]. Despite these promising results, FAPI-PET imaging faces inherent limitations including heterogeneity of FAP expression in cancer-associated fibroblasts, difficulty distinguishing active inflammation from early fibrotic changes (both share transforming growth factor-β activation pathways), and lack of standardized quantification protocols[119]. Moreover, ⁶⁸Ga-FAPI showed no uptake in one patient with biopsy-proven membranous GN in an IgG4-related disease cohort, suggesting variable sensitivity across different GN subtypes[120]. While these advanced PET techniques provide complementary functional and molecular information, renal biopsy remains indispensable for definitive histologic classification, immunofluorescence pattern determination, and electron microscopic ultrastructural analysis that guide specific therapeutic strategies in GN.

It should be acknowledged that the reported cut-off values and AUCs are study-specific and depend on population characteristics, imaging protocols, and statistical modeling. External validation, when available, supports the robustness of these findings; however, most models still require prospective multicenter validation before clinical implementation (Table 4).

Table 4 Clinician-oriented overview linking common clinical questions in glomerulonephritis to imaging modalities and clinical setting¹.

Clinical question	Main clinical aim	Most appropriate imaging modality	Primary clinical endpoint	Clinical setting	Rationale
Acute vs chronic disease?	Disease staging	B-mode ultrasound	Chronicity	Standard of care	First-line tool to assess kidney size, cortical thickness, and corticomedullary differentiation
Renal perfusion impairment?	Functional assessment	Doppler ultrasound; CEUS	Perfusion	Standard of care/adjunct	Evaluation of renal blood flow and microvascular perfusion
Active inflammation vs fibrosis?	Activity vs chronic damage	CEUS; elastography	Activity/fibrosis	Adjunct/emerging	CEUS reflects inflammatory hyperemia; elastography estimates tissue stiffness
Extent of interstitial fibrosis?	Prognostic stratification	Elastography; MRI	Fibrosis	Adjunct	Non-invasive estimation of tissue stiffness and parenchymal remodeling
Tubulointerstitial involvement?	Tissue characterization	Multiparametric MRI	Tissue characterization	Adjunct	Provides complementary functional and structural information
Metabolic or inflammatory activity?	Molecular assessment	PET imaging	Activity	Emerging/research	Detects metabolically active inflammatory tissue
Risk stratification and outcome prediction?	Prognosis	Emerging imaging tools (Radiomics; AI)	Monitoring/prognosis	Research	Quantitative analysis of imaging features for predictive modeling

¹Imaging techniques provide complementary non-invasive information on renal morphology and function but cannot replace histopathological evaluation. Renal biopsy remains the reference standard for definitive diagnosis and should be performed whenever clinically indicated.

AI: Artificial intelligence; CEUS: Contrast-enhanced ultrasound; MRI: Magnetic resonance imaging; PET: Positron emission tomography.

Several key limitations are making it difficult to adopt these techniques in clinical settings: Imaging findings can overlap among different GN subtypes[7,54], which complicates the differential diagnosis process. Early microscopic disease might not show any abnormalities on imaging, and we still do not have standardized protocols or reliable cost-effectiveness data. On top of that, machine learning models need to be validated in diverse populations and with various types of equipment[10,41]. Optimal integration follows a tiered approach: Universal B-mode ultrasound/Doppler for monitoring; selective CEUS and elastography for specific scenarios (LN phenotyping, IgAN fibrosis assessment); multiparametric MRI and PET reserved for research and specialized centers.

Future priorities include prospective studies demonstrating that imaging-guided management improves outcomes, health economics analyses, artificial intelligence validation across diverse settings, and longitudinal cohorts establishing temporal relationships between imaging biomarkers and progression. Renal biopsy remains irreplaceable for diagnosis and treatment selection. Imaging’s strategic value lies in risk stratification, monitoring progression between biopsies, detecting complications, and reducing unnecessary invasive procedures, not replacing histological diagnosis but enhancing clinical decision-making through complementary non-invasive assessment.

CONCLUSION

Imaging in GN serves a complementary rather than definitive diagnostic role. Conventional B-mode ultrasound and Doppler examination provide immediate prognostic information potentially guiding clinical decisions. Advanced techniques occupy specific niches. CEUS differentiates proliferative from nPLN (AUC: 0.81)[36], potentially influencing immunosuppression intensity. Elastography stages IgAN fibrosis[40,44], while multiparametric MRI provides comprehensive functional assessment[93], though standardization remains incomplete[13]. Nuclear medicine applications, particularly ⁶⁸Ga-FAPI PET correlating with fibrosis severity[12,116] and predicting LN treatment response[118], show promise but face accessibility and cost barriers.