Clinical Focus
- Diagnostic Radiology
Academic Appointments
-
Clinical Associate Professor, Radiology
Professional Education
-
Board Certification: American Board of Radiology, Diagnostic Radiology (2016)
-
Residency: Kaiser Permanente Los Angeles Radiology Residency (2015) CA
-
Internship: Kaiser Permanente Los Angeles Internal Medicine Residency (2011) CA
-
Medical Education: Case Western Reserve School of Medicine (2010) OH
-
Fellowship: Duke University Hospital (2016) NC
Clinical Trials
-
Liquid Biopsy With PET/CT Versus PET/CT Alone in Diagnosis of Small Lung Nodules
Recruiting
The purpose of this study is to determine if a liquid biopsy, a method of detecting cancer from a blood draw, combined with a PET/CT scan, a type of radiological scan, is better at determining whether a lung nodule is cancerous when compared to a PET/CT scan alone. A PET/CT scan is already used for diagnosis of lung nodules, but its efficacy is uncertain in nodules 6-20 mm in size. Therefore, the PET/CT will be evaluated for its diagnostic ability in lesions this size alone and in combination with a liquid biopsy. Secondarily, a machine learning model will be created to see if the combination of the PET/CT imaging data and the liquid biopsy data can predict the presence of cancer.
All Publications
-
Implicit screening of abdominal aortic aneurysms among veterans using lumbar spine MRI.
Current problems in diagnostic radiology
2024
Abstract
BACKGROUND: Early detection of abdominal aortic aneurysms (AAAs) is critical given the high morbidity and mortality of a ruptured aneurysm. Screening ultrasound is recommended for men 65 and 75 years of age with a smoking history. However, studies have shown that the rate of ultrasound screening is low and that implicit AAA screening by abdominal imaging studies that were not originally intended for AAA screening can play a major role in AAA detection.OBJECTIVE: The main objective was to evaluate the role of lumbar spine MRIs as an implicit AAA screening study by assessing the detection rate of AAAs in a broader cohort of veterans that included screening and non-screening populations.METHODS: 4085 consecutive lumbar spine MRIs from our institution between 2/2020 and 9/2023 were retrospectively reviewed. Each study was labeled AAA present, AAA not present, or indeterminate by radiologists. The correlation between the presence of AAAs and cardiovascular risk factors was assessed using multinomial logistic regression.RESULTS: AAAs were present in 89 studies (2.2%) from 80 patients (mean age 75.8 (56-93), M:F 10:0) and absent in 3935 cases (96.3%) from 3310 patients (mean age 61.7 (19-100), M:F 9:1). Indeterminate cases (n=61, 1.5%) were mainly due to incomplete visualization (70.5%). Mean AAA size was 3.6cm with most AAAs (n=43) smaller than 3.5cm. Sixteen AAAs were 3.5-3.9cm, 16 between 4 and 4.9cm, and 6 between 5 and 5.9cm. Artifact precluded measurements in 8 cases. Among the AAA-positive cases, 20 had no prior documentation of AAA. Twenty-one patients with AAAs would not have met the criteria for the routine AAA screening ultrasound. Higher rates of hypertension, hyperlipidemia, and smoking were observed for the AAA cohort at 78.8% (OR 2.037, CI 1.160-3.576, P=.013), 82.5% (2.808, 1.543-5.110, P<.001), and 75% (3.340, 1.979-5.638, P<.001), respectively, compared to the matched no-AAA cohort (58.2%, 57.6%, and 50.8%; n=2055).CONCLUSION: Lumbar spine MRI is a valid modality for implicit screening of AAAs.CLINICAL IMPACT: Those interpreting lumbar spine MRIs should be vigilant about assessing for AAAs, especially in men with a history of hypertension, hyperlipidemia, or tobacco smoking.
View details for DOI 10.1067/j.cpradiol.2024.01.022
View details for PubMedID 38246796
-
The Prognostic Value of Qualitative and Quantitative Stress CMR in Patients with Known or Suspected CAD.
JACC. Cardiovascular imaging
2023
Abstract
BACKGROUND: Recent studies suggest that quantitative cardiac magnetic resonance (CMR) may have more accuracy than qualitative CMR in coronary artery disease (CAD) diagnosis. However, the prognostic value of quantitative and qualitative CMR has not been compared systematically.OBJECTIVES: The objective was to conduct a systematic review and meta-analysis assessing the utility of qualitative and quantitative stress CMR in the prognosis of patients with known or suspected CAD.METHODS: A comprehensive search was performed through Embase, Scopus, Web of Science, and Medline. Studies that used qualitative vasodilator CMR or quantitative CMR assessments to compare the prognosis of patients with positive and negative CMR results were extracted. A meta-analysis was then performed to assess: 1) major adverse cardiovascular events (MACE) including cardiac death, nonfatal myocardial infarction (MI), unstable angina, and coronary revascularization; and 2) cardiac hard events defined as the composite of cardiac death and nonfatal MI.RESULTS: Forty-one studies with 38,030 patients were included in this systematic review. MACE occurred significantly more in patients with positive qualitative (HR: 3.86; 95%CI: 3.28-4.54) and quantitative (HR: 4.60; 95%CI: 1.60-13.21) CMR assessments. There was no significant difference between qualitative and quantitative CMR assessments in predicting MACE (P = 0.75). In studies with qualitative CMR assessment, cardiac hard events (OR: 7.21; 95%CI: 4.99-10.41), cardiac death (OR: 5.63; 95%CI: 2.46-12.92), nonfatal MI (OR: 7.46; 95%CI: 3.49-15.96), coronary revascularization (OR: 6.34; 95%CI: 3.42-1.75), and all-cause mortality (HR: 1.66; 95%CI: 1.12-2.47) were higher in patients with positive CMR.CONCLUSIONS: The presence of myocardial ischemia on CMR is associated with worse clinical outcomes in patients with known or suspected CAD. Both qualitative and quantitative stress CMR assessments are helpful tools for predicting clinical outcomes.
View details for DOI 10.1016/j.jcmg.2023.05.025
View details for PubMedID 37632499
-
Performance of alternative manual and automated deep learning segmentation techniques for the prediction of benign and malignant lung nodules.
Journal of medical imaging (Bellingham, Wash.)
2023; 10 (4): 044006
Abstract
We aim to evaluate the performance of radiomic biopsy (RB), best-fit bounding box (BB), and a deep-learning-based segmentation method called no-new-U-Net (nnU-Net), compared to the standard full manual (FM) segmentation method for predicting benign and malignant lung nodules using a computed tomography (CT) radiomic machine learning model.A total of 188 CT scans of lung nodules from 2 institutions were used for our study. One radiologist identified and delineated all 188 lung nodules, whereas a second radiologist segmented a subset (n=20) of these nodules. Both radiologists employed FM and RB segmentation methods. BB segmentations were generated computationally from the FM segmentations. The nnU-Net, a deep-learning-based segmentation method, performed automatic nodule detection and segmentation. The time radiologists took to perform segmentations was recorded. Radiomic features were extracted from each segmentation method, and models to predict benign and malignant lung nodules were developed. The Kruskal-Wallis and DeLong tests were used to compare segmentation times and areas under the curve (AUC), respectively.For the delineation of the FM, RB, and BB segmentations, the two radiologists required a median time (IQR) of 113 (54 to 251.5), 21 (9.25 to 38), and 16 (12 to 64.25) s, respectively (p=0.04). In dataset 1, the mean AUC (95% CI) of the FM, RB, BB, and nnU-Net model were 0.964 (0.96 to 0.968), 0.985 (0.983 to 0.987), 0.961 (0.956 to 0.965), and 0.878 (0.869 to 0.888). In dataset 2, the mean AUC (95% CI) of the FM, RB, BB, and nnU-Net model were 0.717 (0.705 to 0.729), 0.919 (0.913 to 0.924), 0.699 (0.687 to 0.711), and 0.644 (0.632 to 0.657).Radiomic biopsy-based models outperformed FM and BB models in prediction of benign and malignant lung nodules in two independent datasets while deep-learning segmentation-based models performed similarly to FM and BB. RB could be a more efficient segmentation method, but further validation is needed.
View details for DOI 10.1117/1.JMI.10.4.044006
View details for PubMedID 37564098
View details for PubMedCentralID PMC10411216
-
ACR Appropriateness Criteria Workup of Noncerebral Systemic Arterial Embolic Source.
Journal of the American College of Radiology : JACR
2023; 20 (5S): S285-S300
Abstract
Noncerebral systemic arterial embolism, which can originate from cardiac and noncardiac sources, is an important cause of patient morbidity and mortality. When an embolic source dislodges, the resulting embolus can occlude a variety of peripheral and visceral arteries causing ischemia. Characteristic locations for noncerebral arterial occlusion include the upper extremities, abdominal viscera, and lower extremities. Ischemia in these regions can progress to tissue infarction resulting in limb amputation, bowel resection, or nephrectomy. Determining the source of arterial embolism is essential in order to direct treatment decisions. This document reviews the appropriateness category of various imaging procedures available to determine the source of the arterial embolism. The variants included in this document are known arterial occlusion in the upper extremity, lower extremity, mesentery, kidneys, and multiorgan distribution that are suspected to be of embolic etiology. The American College of Radiology Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed annually by a multidisciplinary expert panel. The guideline development and revision include an extensive analysis of current medical literature from peer reviewed journals and the application of well-established methodologies (RAND/UCLA Appropriateness Method and Grading of Recommendations Assessment, Development, and Evaluation or GRADE) to rate the appropriateness of imaging and treatment procedures for specific clinical scenarios. In those instances where evidence is lacking or equivocal, expert opinion may supplement the available evidence to recommend imaging or treatment.
View details for DOI 10.1016/j.jacr.2023.02.005
View details for PubMedID 37236749
-
ACR Appropriateness Criteria Chronic Chest Pain-High Probability of Coronary Artery Disease: 2021 Update.
Journal of the American College of Radiology : JACR
2022; 19 (5S): S1-S18
Abstract
Management of patients with chronic chest pain in the setting of high probability of coronary artery disease (CAD) relies heavily on imaging for determining or excluding presence and severity of myocardial ischemia, hibernation, scarring, and/or the presence, site, and severity of obstructive coronary lesions, as well as course of management and long-term prognosis. In patients with no known ischemic heart disease, imaging is valuable in determining and documenting the presence, extent, and severity of obstructive coronary narrowing and presence of myocardial ischemia. In patients with known ischemic heart disease, imaging findings are important in determining the management of patients with chronic myocardial ischemia and can serve as a decision-making tool for medical therapy, angioplasty, stenting, or surgery. This document summarizes the recent growing body of evidence on various imaging tests and makes recommendations for imaging based on the available data and expert opinion. This document is focused on epicardial CAD and does not discuss the microvascular disease as the cause for CAD. The American College of Radiology Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed annually by a multidisciplinary expert panel. The guideline development and revision include an extensive analysis of current medical literature from peer reviewed journals and the application of well-established methodologies (RAND/UCLA Appropriateness Method and Grading of Recommendations Assessment, Development, and Evaluation or GRADE) to rate the appropriateness of imaging and treatment procedures for specific clinical scenarios. In those instances where evidence is lacking or equivocal, expert opinion may supplement the available evidence to recommend imaging or treatment.
View details for DOI 10.1016/j.jacr.2022.02.021
View details for PubMedID 35550795
-
Preoperative Computed Tomography Angiography Reveals Leaflet-Specific Calcification and Excursion Patterns in Aortic Stenosis.
Circulation. Cardiovascular imaging
1800: CIRCIMAGING121012884
Abstract
BACKGROUND: Computed tomography-based evaluation of aortic stenosis (AS) by calcium scoring does not consider interleaflet differences in leaflet characteristics. Here, we sought to examine the functional implications of these differences.METHODS: We retrospectively reviewed the computed tomography angiograms of 200 male patients with degenerative calcific AS undergoing transcatheter aortic valve replacement and 20 male patients with normal aortic valves. We compared the computed tomography angiography (CTA)-derived aortic valve leaflet calcification load (AVLCCTA), appearance, and systolic leaflet excursion (LEsys) of individual leaflets. We performed computer simulations of normal valves to investigate how interleaflet differences in LEsys affect aortic valve area. We used linear regression to identify predictors of leaflet-specific calcification in patients with AS.RESULTS: In patients with AS, the noncoronary cusp (NCC) carried the greatest AVLCCTA (365.9 [237.3-595.4] Agatston unit), compared to the left coronary cusp (LCC, 278.5 [169.2-478.8] Agatston unit) and the right coronary cusp (RCC, 240.6 [137.3-439.0] Agatston unit; both P<0.001). However, LCC conferred the least LEsys (42.8 [38.8-49.0]) compared to NCC (44.8 [41.1-49.78], P=0.001) and RCC (47.7 [42.0-52.3], P<0.001) and was more often characterized as predominantly thickened (23.5%) compared to NCC (12.5%) and RCC (16.5%). Computer simulations of normal valves revealed greater reductions in aortic valve area following closures of NCC (-32.2 [-38.4 to -25.8]%) and RCC (-35.7 [-40.2 to -32.9]%) than LCC (-24.5 [-28.5 to -18.3]%; both P<0.001). By linear regression, the AVLCCTA of NCC and RCC, but not LCC, predicted LEsys (both P<0.001) in patients with AS. Both ostial occlusion and ostial height of the right coronary artery predicted AVLCCTA, RCC (P=0.005 and P=0.001).CONCLUSIONS: In male patients, the AVLCCTA of NCC and RCC contribute more to AS than that of LCC. LCC's propensity for noncalcific leaflet thickening and worse LEsys, however, should not be underestimated when using calcium scores to assess AS severity.
View details for DOI 10.1161/CIRCIMAGING.121.012884
View details for PubMedID 34915729
-
Dynamic Myocardial Perfusion CT for the Detection of Hemodynamically Significant Coronary Artery Disease.
JACC. Cardiovascular imaging
2021
Abstract
OBJECTIVES: In this international, multicenter study, using third-generation dual-source computed tomography (CT), we investigated the diagnostic performance of dynamic stress CT myocardial perfusion imaging (CT-MPI) in addition to coronary CT angiography (CTA) compared to invasive coronary angiography (ICA) and invasive fractional flow reserve (FFR).BACKGROUND: CT-MPI combined with coronary CTA integrates coronary artery anatomy with inducible myocardial ischemia, showing promising results for the diagnosis of hemodynamically significant coronary artery disease in single-center studies.METHODS: At 9 centers in Europe, Japan, and the United States, 132 patients scheduled for ICA were enrolled; 114 patients successfully completed coronary CTA, adenosine-stress dynamic CT-MPI, and ICA. Invasive FFR was performed in vessels with 25% to 90% stenosis. Data were analyzed by independent core laboratories. For the primary analysis, for each coronary artery the presence of hemodynamically significant obstruction was interpreted by coronary CTA with CT-MPI compared to coronary CTA alone, using an FFR of≤0.80 and angiographic severity as reference. Territorial absolute myocardial blood flow (MBF) and relative MBF were compared using C-statistics.RESULTS: ICA and FFR identified hemodynamically significant stenoses in 74 of 289 coronary vessels (26%). Coronary CTA with≥50% stenosis demonstrated a per-vessel sensitivity, specificity, and accuracy for the detection of hemodynamically significant stenosis of 96% (95%CI: 91-100), 72% (95%CI: 66-78), and 78% (95%CI: 73-83), respectively. Coronary CTA with CT-MPI showed a lower sensitivity (84%; 95%CI: 75-92) but higher specificity (89%; 95%CI: 85-93) and accuracy (88%; 95%CI: 84-92). The areas under the receiver-operating characteristic curve of absolute MBF and relative MBF were 0.79 (95%CI: 0.71-0.86) and 0.82 (95%CI: 0.74-0.88), respectively. The median dose-length product of CT-MPI and coronary CTA were 313 mGy·cm and 138 mGy·cm, respectively.CONCLUSIONS: Dynamic CT-MPI offers incremental diagnostic value over coronary CTA alone for the identification of hemodynamically significant coronary artery disease. Generalized results from this multicenter study encourage broader consideration of dynamic CT-MPI in clinical practice. (Dynamic Stress Perfusion CT for Detection of Inducible Myocardial Ischemia [SPECIFIC]; NCT02810795).
View details for DOI 10.1016/j.jcmg.2021.07.021
View details for PubMedID 34538630
-
Machine Learning Radiomics Model for Early Identification of Small-Cell Lung Cancer on Computed Tomography Scans.
JCO clinical cancer informatics
2021; 5: 746-757
Abstract
PURPOSE: Small-cell lung cancer (SCLC) is the deadliest form of lung cancer, partly because of its short doubling time. Delays in imaging identification and diagnosis of nodules create a risk for stage migration. The purpose of our study was to determine if a machine learning radiomics model can detect SCLC on computed tomography (CT) among all nodules at least 1 cm in size.MATERIALS AND METHODS: Computed tomography scans from a single institution were selected and resampled to 1 * 1 * 1 mm. Studies were divided into SCLC and other scans comprising benign, adenocarcinoma, and squamous cell carcinoma that were segregated into group A (noncontrast scans) and group B (contrast-enhanced scans). Four machine learning classification models, support vector classifier, random forest (RF), XGBoost, and logistic regression, were used to generate radiomic models using 59 quantitative first-order and texture Imaging Biomarker Standardization Initiative compliant PyRadiomics features, which were found to be robust between two segmenters with minimum Redundancy Maximum Relevance feature selection within each leave-one-out-cross-validation to avoid overfitting. The performance was evaluated using a receiver operating characteristic curve. A final model was created using the RF classifier and aggregate minimum Redundancy Maximum Relevance to determine feature importance.RESULTS: A total of 103 studies were included in the analysis. The area under the receiver operating characteristic curve for RF, support vector classifier, XGBoost, and logistic regression was 0.81, 0.77, 0.84, and 0.84 in group A, and 0.88, 0.87, 0.85, and 0.81 in group B, respectively. Nine radiomic features in group A and 14 radiomic features in group B were predictive of SCLC. Six radiomic features overlapped between groups A and B.CONCLUSION: A machine learning radiomics model may help differentiate SCLC from other lung lesions.
View details for DOI 10.1200/CCI.21.00021
View details for PubMedID 34264747
-
ACR Appropriateness Criteria Infective Endocarditis.
Journal of the American College of Radiology : JACR
2021; 18 (5S): S52–S61
Abstract
Infective endocarditis can involve a normal, abnormal, or prosthetic cardiac valve. The diagnosis is typically made clinically with persistently positive blood cultures, characteristic signs and symptoms, and echocardiographic evidence of valvular vegetations or valvular complications such as abscess, dehiscence, or new regurgitation. Imaging plays an important role in the initial diagnosis of infective endocarditis, identifying complications, prognostication, and informing the next steps in therapy. This document outlines the initial imaging appropriateness of a patient with suspected infective endocarditis and for additional imaging in a patient with known or suspected infective endocarditis. The American College of Radiology Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed annually by a multidisciplinary expert panel. The guideline development and revision include an extensive analysis of current medical literature from peer reviewed journals and the application of well-established methodologies (RAND/UCLA Appropriateness Method and Grading of Recommendations Assessment, Development, and Evaluation or GRADE) to rate the appropriateness of imaging and treatment procedures for specific clinical scenarios. In those instances where evidence is lacking or equivocal, expert opinion may supplement the available evidence to recommend imaging or treatment.
View details for DOI 10.1016/j.jacr.2021.01.010
View details for PubMedID 33958118
-
CovXR: Automated Detection of COVID-19 Pneumonia in Chest X-Rays through Machine Learning
IEEE. 2021
View details for DOI 10.1109/SSCI50451.2021.9659919
View details for Web of Science ID 000824464300100