Feasibility of contrast-enhanced ultrasound as a screening tool for patients at high risk for kidney malignancies

Introduction

Some rare, syndromic genetic diseases [e.g., von Hippel-Lindau (VHL), tuberous sclerosis (TS)] predispose patients to kidney malignancies, necessitating imaging surveillance with techniques such as contrast-enhanced computed tomography (CT), magnetic resonance imaging (MRI), and non-contrast ultrasound (1). Contrast-enhanced CT and MRI accurately detect lesions, but repeated iodinated contrast exposure for CT scans can hasten kidney function decline and certain gadolinium contrast agents for MRI can increase the risk of nephrogenic sclerosing fibrosis among individuals with moderate to severe chronic kidney dysfunction. While non-contrast-enhanced CT and MRI can be used to visualize the kidneys, the lack of microvascular enhancement compromises accuracy, and CT exposes patients to radiation. Alternatively, contrast-enhanced ultrasound (CEUS) using Definity^® microbubble contrast captures vasculature and can be used to characterize kidney and liver lesions (2) without the risks associated with CT and MRI contrast agents or radiation exposure. To date, CEUS has been used predominantly as a diagnostic rather than screening tool (3). In this study, we hypothesized that CEUS would be as sensitive but not as specific as CT/MRI making it a test worthy of investigation as a screening tool for kidney lesions in individuals with predisposing risk factors for kidney malignancy.

Methods

Study cohort and objective assessment

Adults with VHL, TS, Birt-Hogg-Dubé (BHD), or hereditary leiomyomatosis and renal cell cancer (HLRCC) were eligible. Sonographers with at least 5 years of clinical experience performed bilateral ultrasound in B-mode and CEUS mode with bolus-dosed Definity^® contrast (Lantheus, Bedford, MA, USA). Clips were taken in transverse and longitudinal planes at steady state of enhancement. Three blinded radiologists (L.B., E.L. and K.O.), each with experience in CEUS interpretation (4, 7, and 6 years of experience, respectively), reviewed, in scrambled order, B-mode only (58 clips) or B-mode/CEUS combined (116 clips) and answered standardized questions about lesion detection (yes/no) and malignancy status (yes/no/unsure). We used kappa statistics (κ) to compare the reader assessment of CEUS imaging to the surrogate gold standard of cross-sectional contrast CT or MRI within 3 months of CEUS. We calculated absolute κ for binary outcomes (lesion identification), where readers either agreed or disagreed as to lesion presence (yes or no), and weighted κ for malignancy status (4-6) (yes, no, unsure) to reflect the degree of disagreement between raters (larger disagreement between yes and no than between yes and unsure, for example). For both outcomes, we used the means of κ for all individual reader pairs and determined the overall agreement between readers and the gold standard by averaging the κ for all reader-gold standard pairs. We also calculated the sensitivity and specificity of B-mode and, separately, combined B-mode/CEUS for lesion detection and malignancy status (vs. gold standard imaging). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Institutional Review Board (IRB) at the University of North Carolina at Chapel Hill (No. 17-2055). Informed consent was obtained from all participants before they enrolled in the study.

Software backbone and radiologist reviewers’ pipeline

The software platform developed for this study was designed to enable standardized, blinded radiologist evaluation of kidney ultrasound imaging data. The system backbone consisted of an image processing pipeline, a custom graphical user interface (GUI), and an automated data/entry capture framework, as shown in Figure 1A.

Figure 1 Development and implementation of the ultrasound image review platform and radiologist assessment workflow. (A) Overview of the data curation and software backbone illustrating the imaging data workflow, including DICOM offloading from ultrasound systems, conversion to AVI format, selection of optimal B-mode and contrast-enhanced scans, dataset organization and anonymization, GUI development, standalone software packaging, distribution to independent radiologist readers, and downstream statistical analysis of reader assessments. (B) Representative screenshots of the implemented GUI demonstrating folder and scan selection, structured assessment questionnaire, synchronized video playback with temporal navigation, and integrated scoring interface for standardized radiologist evaluation. DICOM, Digital Imaging and Communications in Medicine; GUI, graphical user interface.

Kidney ultrasound examinations were originally acquired on clinical ultrasound systems and stored in Digital Imaging and Communications in Medicine (DICOM) format and later converted to AVI format. For each subject, representative high-quality scan sequences were compiled for both B-mode and CEUS imaging modalities. To ensure full reader blinding, all imaging folders and scan identifiers were anonymized through randomly generated naming conventions.

We implemented the software in Python environment (available at https://github.com/ArianAzg/Kidney_ultrasound_assessment_software), with the Tkinter library serving as the primary framework for GUI development. The application was compiled and distributed as a standalone Windows executable file, allowing deployment across clinical workstations. The GUI was engineered to provide an intuitive workflow consistent with radiologic review practices. When using the program, readers selected anonymized study folders and loaded available scans for assessment. Videos were presented using standardized naming conventions encoding laterality, modality, and acquisition plane, enabling efficient navigation while preserving blinding.

The interface supported simultaneous visualization of multiple videos, allowing radiologists to compare complementary views during interpretation. Scans were equipped with an interactive sliding time bar to facilitate frame-level navigation. Study-specific evaluation questions, predefined by the investigative team, were embedded directly within the software environment (Figure 1B). Readers completed structured assessments for each kidney using all available scans.

Reader inputs were captured through an automated logging system integrated within the application. Each record corresponded to a unique scan evaluation and retained anonymized identifiers to ensure blinded downstream analysis. This structured data capture framework enabled efficient aggregation of reader responses and supported subsequent statistical analyses, including inter-reader agreement.

Results

Participants had a mean age of 40 years, 53% (8/15) were female, 20% (3/15) were Black, and 67% (10/15) had VHL (Table 1). Fifteen participants (29 kidneys) underwent CT/MRI (gold standard) and B-mode/CEUS imaging, an example of which is shown in Figure 2. CT/MRI diagnoses were obtained from the clinical charts. Of the 29 kidneys, eight had CT/MRI-detected lesions; three of those kidneys had at least one malignant lesion. The sizes of the lesions as measured from MRI ranges from 0.3 to 3.5 cm (median size: 2.6 cm). All three readers detected the same six kidneys as having lesions (κ =1) using either B-mode or combined B-mode/CEUS. The overall agreement between ultrasound readers and the gold standard for lesion detection (contrast CT or MRI) was κ =0.81 [95% confidence interval (CI): 0.49–1.0] for both B-mode alone and B-mode/CEUS. For lesion detection, B-mode/CEUS had a specificity of 100% (21/21; 95% CI: 100–100%) and sensitivity of 75% (6/8; 95% CI: 40–100%).

Figure 2 CT, B-Mode, and CEUS for left kidney; CT: (A) axial plane, (B) coronal plane, (C) sagittal plane, B-Mode: (D) longitudinal plane, and CEUS: (E) longitudinal plane. All images depict the same left kidney across different imaging modalities. Images demonstrate a simple, non-enhancing partially exophytic cyst from the upper pole (red arrows), a complex, internally enhancing lesion involving the interpolar region (blue arrows), and a complex, internally enhancing lesion involving the lower pole (green arrows). CEUS, contrast-enhanced ultrasound; CT, computed tomography.

For malignancy status, the overall kappa for inter-reader agreement was 0.64 for B-mode and 0.66 for B-mode/CEUS. Among the 6 kidneys with lesions detected by B-mode/CEUS, the overall agreement between ultrasound readers and the gold standard for malignancy status (contrast CT or MRI) was κ =0.25 for B-mode and κ =0.35 for B-mode/CEUS. While lesion detection was comparable, B-mode alone did not correctly identify any malignant lesions, whereas B-mode/CEUS improved sensitivity for malignancy assessment. For malignancy status, B-mode/CEUS had a sensitivity of 100% (3/3; 95% CI: 0–100%) and specificity of 88% (23/26; 95% CI: 75–100%). The lesions from two different kidneys which were missed with B-mode/CEUS, but detected with CT/MRI, were identified as non-malignant.

Discussion

This pilot study suggests that B-mode/CEUS can be as a complementary imaging technique that may be used to reduce the frequency of CT/MRI for surveillance among patients at high-risk for developing kidney malignancies. We demonstrated that CEUS was feasible and had perfect inter-reader agreement for lesion detection and fair inter-reader agreement for malignancy status. The modest agreement may be attributed to intrinsic features of CEUS in fully characterizing lesion vascularity compared to cross-sectional imaging, particularly in small or indeterminate lesions. Variability in reader experience may further contribute. From a clinical perspective, false positive CEUS findings would trigger follow-up imaging; however, this aligns with a conservative diagnostic approach prioritizing sensitivity, where CEUS functions as a complementary imaging technique. MRI/CT provide cross-sectional views and greater resolution than CEUS and, as expected, detected more lesions than CEUS. However, all lesions not detected by B-mode/CEUS were benign. B-mode/CEUS misclassified three MRI/CT designated benign lesions as malignant. While this misclassification could be disconcerting to patients and clinicians, it would be unlikely to have important clinical consequences as the missed lesions were small (≤2 cm), non-enhancing, cystic lesions (including a 3 mm simple cyst and subcentimeter hemorrhagic/proteinaceous cysts), and lesions would be further investigated by CT/MRI. Overall, our findings suggest that B-mode/CEUS may be a viable screening tool for malignant kidney lesions among people with conditions predisposing them to kidney malignancy. These proportions should be interpreted cautiously given the limited sample size. There are several limitations in our study including the lack of pathologic confirmation since lesions were typically managed noninvasively. Another limitation is that CT/MRI were clinical protocols. Larger studies are needed to establish the clinical utility of CEUS in routine screening.

Conclusions

With 100% sensitivity for malignant lesions and high reliability across readers, CEUS is a powerful tool for detecting kidney cancer in high-risk populations. By integrating CEUS into screening workflows, clinicians may be able to reduce the frequency of more invasive or expensive cross-sectional imaging like CT and MRI. While promising, these findings should be confirmed in larger studies.

Acknowledgments

The authors gratefully acknowledge the support of Lantheus Medical Imaging grant CG#19001 which provided essential resources for this research. The project described was supported by the National Center for Advancing Translational Sciences (NCATS), National Institutes of Health (grant award No. UM1TR004406). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Footnote

Peer Review File: Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0181/prf

Funding: This work was supported by Lantheus Medical Imaging grant CG#19001 and the National Center for Advancing Translational Sciences (NCATS), National Institutes of Health (grant award No. UM1TR004406).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0181/coif). K.A.J. reports that grants (NCSU CMI and NCDRC) were awarded for research and allocated to K.A.J.’s department. K.A.J. also reports receiving a travel award to attend the ABRCMS conference, as well as compensation for facilitating leadership modules for the UNC Center for Faculty Excellence. R.W.W. reports that the postdoctoral appointment was supported by IRACDA from September 2022 through February 2025 (grant 2K12GM000678-24). L.B. reports receiving speaking fees from a Bayer-sponsored meeting in September 2025. E.H.C. was compensated for serving as a speaker in a seminar series and for compiling imaging training materials for Lantheus Medical Imaging. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Institutional Review Board (IRB) at the University of North Carolina at Chapel Hill (No. 17-2055). Informed consent was obtained from all participants before they enrolled in the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Louise M, Binderup M, Smerdel M, Borgwadt L, et al. von Hippel-Lindau disease: Updated guideline for diagnosis and surveillance. Eur J Med Genet 2022;65:104538. [Crossref] [PubMed]
2. Spiesecke P, Thiemann J, Conen P, et al. Contrast enhanced ultrasound of cystic renal lesions, from diagnosis up to treatment. Clin Hemorheol Microcirc 2024;88:S21-33. [Crossref] [PubMed]
3. Zbroja M, Kuczyńska M, Drelich K, et al. Contrast-Enhanced Ultrasound in the Diagnosis of Solid Renal Lesions. J Clin Med 2024;13:3821. [Crossref] [PubMed]
4. Cohen J. A Coefficient of Agreement for Nominal Scales. Educational and Psychological Measurement 1960;20:37-46.
5. Cohen J. Weighted kappa: Nominal scale agreement provision for scaled disagreement or partial credit. Psychological Bulletin 1968;70:213-20.
6. Landis JR, Koch GG. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 1977;33:363-74.