Kline, Timothy L.

Background: Total kidney volume (TKV), a prognostic biomarker in ADPKD, is used to stratify high risk individuals. eGFR measures may carry significant error in ADPKD compared to measured GFR(mGFR). This retrospective study investigates the association between height adjusted TKV (htTKV) and eGFR derived from race-free CKD-EPI equations utilizing creatinine and cystatin C. Methods: eGFR was calculated using 3 race-free CKD-EPI formulas: creatinine-based (CKD-EPI creatinine), cystatin C (CKD-EPI cys), and creatinine-cystatin C (CKD-EPI creatinine-cystatin C) equations on 697 Mayo Clinic ADPKD patients. Correlations between eGFR and TKV measures were estimated and stratified by MIC class risk groups and CKD. Linear regression models were computed to predict ln(htTKV) from the 3 eGFR data sets. The eGFR measures were compared to iothalamate-based mGFR in a subset of patients (n=50). Results: A correlation between eGFR measures and htTKV (Pearson correlation r of -0.53 to -0.54), varied by MIC stratification. Pearson correlation was -0.47 to -0.48 for MIC 1A and 1B, -0.67 to -0.71 for MIC 1C, and -0.67 to -0.68 for MIC 1D and 1E. All equations were significantly associated with htTKV levels, with similar adjusted beta values (95% CI) of -0.15 (-0.17 to -0.13), -0.14 (-0.16 to -0.12), -0.15 (-0.16 to -0.13), for CKD-EPI creatinine, CKD-EPI cystatin C, and CKD-EPI creatinine-cystatin C, respectively. Compared to mGFR, the CKD-EPI creatinine-cystatin C equation was the least biased (mean -0.86 (95% CI: -3.09 to 1.37)) versus CKD-EPI creatinine (2.31 (-0.73 to 5.34)) or CKD-EPI cystatin C (-3.54 (-5.86 to -1.21)), and most precise, with 1-P30 of 16% (95% CI 7.2% to 29.1%), compared to 18% (8.6% to 31.4%) for CKD-EPI creatinine, or 20% (10.0% to 33.7%) for CKD-EPI cystatin C. Conclusion: A moderate correlation between eGFR and htTKV associated with MIC stratification. Associations between htTKV and all 3 CKD-EPI equations were similar. CKDEPIcrcys tended to be the least biased and most precise equation when compared to mGFR and may be a reasonable alternative.

Key Points In a cohort of children, 79% of families with kidney cysts had a monogenic diagnosis. Resolved cases were more likely to have a family history of kidney cysts and unresolved cases more often had unilateral cysts. Broad genetic testing revealed genetic diversity and informed prognosis and clinical management in childhood kidney cysts. Background Pediatric kidney cysts may indicate an underlying genetic disorder, yet the full spectrum of causes remains incompletely defined. With advances in genetic testing enabling broader evaluation, this study assessed the diagnostic utility of comprehensive genetic testing in a broad pediatric cohort with kidney cysts and characterized the underlying etiological diversity. Methods This observational cohort study included patients younger than 18 years enrolled between January 2020 and June 2024 at a single tertiary center. Genetic testing used targeted multigene or custom curated exome/genome sequencing panels, with segregation analysis when available. Eligible participants had at least two cysts without family history, one cyst with positive family history, or enlarged echogenic kidneys on prenatal ultrasound; those with congenital anomalies of the kidney and urinary tract associated with cysts were excluded. Clinical presentation was categorized as symptomatic, incidental, prenatal, or family screening. Primary outcomes were diagnostic yield and distribution of pathogenic variants. Results Among 109 patients (median age 7.6 years, 53% female), genetic testing identified a definitive diagnosis in 81 of 100 tested patients (81%) from 72 families (79%; 14 disorders). PKD1 variants were most common (45%), while PKD2 accounted for 7%. Other causes included HNF1B or 17q12 deletions (13%), minor autosomal dominant polycystic kidney disease genes (11%; GANAB, NEK8, IFT140 ), monoallelic PKHD1 (3%), and biallelic PKHD1 (8%), while syndromic ciliopathy genes accounted for 5%. A positive family history of cystic kidney disease was more common among patients with a genetic diagnosis (54% versus 11%; P = 0.008). Patients without an identified genetic diagnosis more often had unilateral cysts (26% versus 4%; P = 0.53). Diagnosis by clinical symptoms was the most genetically diverse category. Conclusions Comprehensive genetic testing in pediatric kidney cysts achieved high diagnostic yield in a selected cohort and identified diverse causes beyond major autosomal dominant polycystic kidney disease/autosomal recessive polycystic kidney disease genes, supporting early evaluation to improve diagnostic accuracy, inform prognosis, and guide management, even in the absence of family history.

What was done? A review of artificial intelligence (AI) applications for the imaging of uterine fibroids, endometriosis, and adenomyosis. What was found? AI models can assist with the recognition, segmentation, and localization of uterine fibroids, and the differentiation of benign fibroids and sarcomas. Models can aid in the diagnosis of adenomyosis and endometriosis, and the prediction of the impact of endometriosis on fertility. What the findings mean? Deployed thoughtfully, AI tools could reduce variability, shorten read times, and add objective measurements to routine care. Studies evaluating these models are limited by single-institution designs and continued reliance on expert sonologists and radiologists.

Abstract Medical imaging is undergoing a transformation driven by the advent of new, highly effective, machine learning techniques paired with increases in computational capabilities (Cheng et al. 2021; Gilson et al. 2023; Almeida et al. 2024; Krishna et al. 2024). These advanced algorithms have the potential to improve disease detection, diagnosis, prognosis, and treatment outcomes. However, the complexity of machine learning models, the large amounts of curated and annotated data required by some methods, and the potential for bias and error make it challenging for individuals to safely and effectively leverage these methods (Lin et al. 2024; Guo et al. 2024; Xu et al. 2024; Linguraru et al. 2024; Wood et al. 2019). To address these challenges, the American Association of Physicists in Medicine (AAPM), American College of Radiology (ACR), Radiological Society of North America (RSNA), and Society for Imaging Informatics in Medicine (SIIM) have worked together to develop a syllabus detailing a recommended set of competencies for medical imaging professionals interacting with these systems. This guide is aimed at four different personas: users of AI systems, purchasers of AI systems, individuals who provide clinical expertise during the development of AI systems (“clinical collaborators”), and developers of AI systems.1 This is a syllabus, not a curriculum, and is intentional in this scope. Recognizing that individuals may benefit from different presentations of the same material, this work enumerates a series of relevant competencies but does not prescribe, nor offer, a method of instruction (Schuur, Rezazade Mehrizi, and Ranschaert 2021; Garin et al. 2023). By addressing the task‐specific demands of each role, this guide will enable medical imaging professionals to utilize machine learning systems more safely and effectively, ultimately improving patient care and outcomes.

Abstract Radiological protocol selection is a critical but time-consuming step in clinical workflow, requiring radiologists to match patient indications with the appropriate MRI or CT protocol. Manual selection can be prone to delays or potential errors, and automated approaches must contend with substantial data imbalance, site-specific variation, and evolving nomenclature. We investigated whether a large language model (LLM) can support reliable protocol selection at scale and whether retrieval-augmented generation (RAG) offers operational advantages over direct fine-tuning. Using 498,228 patient reports collected across three Mayo Clinic sites (Arizona, Florida, and Rochester) spanning six radiological divisions, we trained site-specific Llama 3.2 3B models for use with and without retrieval augmentation. Division-scoped Facebook AI Similarity Search (FAISS) indexes constructed from procedure and diagnosis text were used to supply contextual evidence in the RAG framework. Both fine-tuned and RAG-augmented models achieved strong performance across sites, with F1 scores of 0.88–0.90. RAG matched or modestly trailed direct fine-tuning overall but delivered consistent gains in specific divisions (e.g., musculoskeletal imaging). Importantly, the RAG model introduced abstention behavior (Not Enough Information), which concentrated in linguistically diverse divisions and provided an interpretable signal of uncertainty. These findings suggest that RAG-based models are viable for division-scoped protocol selection and offer practical advantages. Retrieval indexes can be refreshed far more easily and with fewer resources than retraining LLMs, enabling continual adaptation to evolving clinical workflows. Future prospective deployment will evaluate real-time accuracy, agreement with practitioners, and the role of abstention as a safety mechanism in clinical decision support.