Clinical Focus

  • Body Imaging

Academic Appointments

  • Clinical Assistant Professor, Radiology

Professional Education

  • Residency: University of Illinois at Chicago Radiology Residency (2016) IL
  • Fellowship: Stanford University Radiology Fellowships (2017) CA
  • Board Certification: American Board of Radiology, Diagnostic Radiology (2017)
  • Medical Education: University of Illinois College of Medicine Office of the Registrar (2011) IL
  • Internship: Presence Saint Joseph Hospital (2012) IL

All Publications

  • Diagnostic performance of hypoechoic perinephric fat as a predictor of prediabetes and diabetes. Abdominal radiology (New York) Shen, L., Tse, J. R., Negrete, L. M., Shon, A., Yoon, L., Liang, T., Kamaya, A. 2022


    To evaluate prevalence and predictive value of hypoechoic perinephric fat (HPF) in patients with prediabetes and diabetes compared to non-diabetics.Of 240 patients with renal ultrasound and hemoglobin A1c (HbA1c) measurements, 114 patients had either prediabetes (HbA1c 5.7-6.4%) or diabetes (HbA1c ≥ 6.5%), and 126 patients did not. Two radiologists (blinded to diagnosis) reviewed images and discrepancies were resolved by a third. Inter-reader agreement was compared using free-marginal kappa and intraclass correlation coefficient. Fisher's exact test, Mann-Whitney test, multivariable logistic regression, and Spearman's rank correlation test with two-tailed p < 0.05 were used to determine statistical significance.HPF was exclusively identified in prediabetic and diabetic patients with a prevalence of 23% (vs 0%; p < 0.001). Identification of HPF had almost perfect inter-reader agreement (k = 0.94) and was statistically significant (p = 0.034) while controlling for body mass index (BMI) and estimated glomerular filtration rate in multivariable analysis. HPF had extremely high specificity and positive predictive value (100% for both) in patients with prediabetes and diabetes although it was not a sensitive finding (23% sensitivity). In patients with prediabetes and diabetes, those with HPF were statistically significantly more likely to have chronic kidney disease (CKD) (p = 0.003). There was no statistically significant difference in BMI, stages of CKD, and types of diabetes.Hypoechoic perirenal fat has almost perfect inter-reader agreement and is highly specific for and predictive of prediabetes and diabetes. Its presence may also help identify those with chronic kidney disease among prediabetic and diabetic patients.

    View details for DOI 10.1007/s00261-022-03763-3

    View details for PubMedID 36480029

  • Automated Identification and Measurement Extraction of Pancreatic Cystic Lesions from Free-Text Radiology Reports Using Natural Language Processing. Radiology. Artificial intelligence Yamashita, R., Bird, K., Cheung, P. Y., Decker, J. H., Flory, M. N., Goff, D., Morimoto, L. N., Shon, A., Wentland, A. L., Rubin, D. L., Desser, T. S. 2022; 4 (2): e210092


    Purpose: To automatically identify a cohort of patients with pancreatic cystic lesions (PCLs) and extract PCL measurements from historical CT and MRI reports using natural language processing (NLP) and a question answering system.Materials and Methods: Institutional review board approval was obtained for this retrospective Health Insurance Portability and Accountability Act-compliant study, and the requirement to obtain informed consent was waived. A cohort of free-text CT and MRI reports generated between January 1991 and July 2019 that covered the pancreatic region were identified. A PCL identification model was developed by modifying a rule-based information extraction model; measurement extraction was performed using a state-of-the-art question answering system. The system's performance was evaluated against radiologists' annotations.Results: For this study, 430426 free-text radiology reports from 199783 unique patients were identified. The NLP model for identifying PCL was applied to 1000 test samples. The interobserver agreement between the model and two radiologists was almost perfect (Fleiss kappa = 0.951), and the false-positive rate and true-positive rate were 3.0% and 98.2%, respectively, against consensus of radiologists' annotations as ground truths. The overall accuracy and Lin concordance correlation coefficient for measurement extraction were 0.958 and 0.874, respectively, against radiologists' annotations as ground truths.Conclusion: An NLP-based system was developed that identifies patients with PCLs and extracts measurements from a large single-institution archive of free-text radiology reports. This approach may prove valuable to study the natural history and potential risks of PCLs and can be applied to many other use cases.Keywords: Informatics, Abdomen/GI, Pancreas, Cysts, Computer Applications-General (Informatics), Named Entity Recognition Supplemental material is available for this article. © RSNA, 2022See also commentary by Horii in this issue.

    View details for DOI 10.1148/ryai.210092

    View details for PubMedID 35391762