Background/Purpose: Salivary gland ultrasound (SGUS) has the potential to inform diagnosis and progression of Sjogren’s disease, a systemic autoimmune disease which results in progressive inflammation and loss of exocrine gland function. SGUS has primarily been performed and read by rheumatologists in Europe. Despite its potential benefit, SGUS is not widely performed in the United States. Radiology departments have the resources to perform SGUS, however there remains little referral. To assess the feasibility of radiology performed SGUS in the United States, we sought to review the implementation and outcomes of a standardized SGUS protocol at a large urban academic radiology department in the United States which includes a large number of radiologists with diverse experience.

Methods: A standardized SGUS protocol was developed by the radiology department in consultation with rheumatology. SGUS were performed and read by two groups, a general radiology practice, and an academic practice. A retrospective analysis of these studies was performed. SGUS between August 2021, when the protocol was instituted, and February 2025 were identified within the radiology information system stored within a web-based data lake. Meta data within the image DICOM header was used to calculate study metrics such as time to complete study, number of images within study, and storage size of study. A large language model (LLM) was used to evaluate the consistency of the reports with the standardized template and to extract OMERACT score and gland size from the reports. Outcomes for patients with suspected lymphoma were reviewed.

Results: A total of 811 SGUS were performed over the 41-month period. The mean age of the patients was 54 ± 15 years with 89% (721/811) female. The study was ordered by 58 different departments with 78% (632/811) of the orders coming from a Rheumatology practice. Studies were read by 53 different radiologists, 55% of which were in the academic group. The academic group read 56% (458/811) of the studies. Exams took a median of 15 minutes (IQR = 12 min) to complete. Each exam contained an average of 11.8 cines and 84 still images and a median size of 274 MB (IQR = 223 MB). Analysis of the report text by the LLM demonstrated 64% (519/811) to be consistent with the template. Of the inconsistent reports, 35% (99/292) were from the academic group. Incomplete scoring of all the glands (at least one gland not given an OMERACT score) was found in 13% (103/811) of the reports of which 71% (73/103) were from the academic group. Incomplete measurements of the glands (at least one dimension not reported) were found in 23% (188/811) of which 79% (149/188) were from the academic group.No association was found between salivary gland volume and OMERACT score. Only 0.5% (4/811) of studies had a recommendation for a gland biopsy due to concern for lymphoma. Of these four, only one was positive for lymphoma, and was likely a recurrence as the patient had a prior history of lymphoma.

Conclusion: SGUS can be reasonably implemented by both general and academic radiologists with minimal time and storage burden. Radiologists create consistent reports which can be used for subsequent cohort analysis.

Salivary gland ultrasound study characteristics. Differences in mean right parotid volume according to maximum OMERACT score calculated via ANOVA. SGUS: salivary gland ultrasound.

Mean salivary gland volume by OMERACT score. No significant difference identified.

Distribution of right parotid volume by OMERACT score. No significant difference identified.

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/salivary-gland-ultrasounds-can-be-read-by-general-radiologists-a-review-of-4-years-of-reports-from-an-urban-academic-radiology-department/