About

Susan P. Weinstein, M.D., is a Professor of Radiology at the Hospital of the University of Pennsylvania and holds active staff positions in the Department of Medical Imaging at Penn Presbyterian Medical Center. She is also an Attending Staff member in the Department of Radiology at the Hospital of the University of Pennsylvania. Dr. Weinstein serves as the Director of the UPHS Breast Imaging Network Practice Optimization Team (NPOT) within Penn Medicine and is involved in strategic projects as Associate Chair for the Radiology Network at the University of Pennsylvania. Her research expertise focuses on breast cancer screening and diagnosis, with particular emphasis on breast MRI. Dr. Weinstein has contributed to numerous studies related to digital breast tomosynthesis, breast density estimation, and the effectiveness of various breast imaging modalities. Her clinical work encompasses breast cancer screening, diagnosis, and image-guided interventions. She has been actively involved in advancing breast imaging practices and improving screening outcomes through her research and clinical leadership.

Research topics

• Internal medicine
• Medicine
• Medical physics
• Radiology

Selected publications

• ACR Appropriateness Criteria® Breast Imaging During Lactation

  Journal of the American College of Radiology · 2026-03-01

  articleOpen access
  PublisherOA PDFDOI

• ACR Appropriateness Criteria® Imaging of Ductal Carcinoma in Situ (DCIS)

  Journal of the American College of Radiology · 2025-05-01

  article
  PublisherDOI

• ACR Appropriateness Criteria® Imaging of Invasive Breast Cancer

  UNC Libraries · 2025-06-02

  articleOpen access
  PublisherDOI

• ACR Appropriateness Criteria® Supplemental Breast Cancer Screening Based on Breast Density: 2024 Update

  Journal of the American College of Radiology · 2025-05-01 · 12 citations

  article
  PublisherDOI

• Convolutional neural network model observers discount signal-like anatomical structures during search in virtual digital breast tomosynthesis phantoms

  Journal of Medical Imaging · 2025-10-16

  article

  PurposeWe aim to assess the perceptual tasks in which convolutional neural networks (CNNs) might be better tools than commonly used linear model observers (LMOs) to evaluate medical image quality.ApproachWe compared the LMOs (channelized Hotelling [CHO] and frequency convolution channels observers [FCO]) and CNN detection accuracies for tasks with a few possible signal locations (location known exactly) and for the search for mass and microcalcification signals embedded in 2D/3D breast tomosynthesis phantoms. We also compared the LMOs and CNN accuracies to those of radiologists in the search tasks. We analyzed radiologists’ eye position to assess whether they fixate longer at locations considered suspicious by the LMOs or those by the CNN.ResultsLMOs resulted in similar detection accuracies [area under the receiver operating characteristic curve (AUC)] to the CNN for tasks with up to 100 signal locations but lower accuracies in the search task for microcalcification and mass 3D images. Radiologists’ AUC was significantly higher (p<1e−4) than that of LMOs for the microcalcification 2D search (CHO, FCO) and 3D mass search (p<0.05, CHO) but was not higher than the CNN’s AUC. For both signal types, radiologists fixated longer on the locations of the highest response scores of the CNN than those of the LMOs but only reached statistical significance for the mass (masses: p=0.009 versus CHO and p=0.004 versus FCO)ConclusionWe show that CNNs are a more suitable model observer for search tasks. Like radiologists but not traditional LMOs, CNNs can discount false positives arising from anatomical backgrounds.

  PublisherDOI

• miRNA panel from HER2+ and CD24+ plasma extracellular vesicle subpopulations as biomarkers of early-stage breast cancer

  Breast Cancer Research · 2025-05-22 · 9 citations

  articleOpen access

  BACKGROUND: Mammography screening has improved early breast cancer detection, leading to reduced mortality and lower rates of advanced breast cancer. However, mammography has a high false positive rate that results in over a million invasive breast biopsies of benign lesions in the US each year. Therefore, there is a need for noninvasive, blood-based diagnostics that can accurately assess risk of malignancy for women with indeterminate lesions identified by mammography, such as BI-RADS category 4 breast lesions. The aim of this study is to identify biomarkers from multiplexed extracellular vesicle liquid biopsy that can accurately classify mammographically detected BI-RADS 4 lesions. METHODS: We analyzed plasma from 113 prospectively enrolled subjects with BI-RADS 4 breast lesions, including 86 women with benign lesions and 27 women with malignant lesions (including 12 with stage I invasive carcinoma and 14 with ductal carcinoma in situ). None of the invasive carcinomas were metastatic. From each plasma sample, we used track etched magnetic nanopore technology to separately isolate HER2 and CD24 expressing extracellular vesicles (EVs) and measured their miRNA cargo using next-generation sequencing. We evaluated the performance of EV-miRNA biomarkers for classifying malignancy and applied LASSO classification to identify a panel of four complementary EV miRNA biomarkers that we validated by qPCR. RESULTS: We identified 19 differentially enriched miRNA from HER2+ EVs and 11 differentially enriched miRNA from CD24+ EVs of women with malignant lesions compared to benign lesions. We observed individual miRNA with an AUC of up to 0.87 for miR-340-5p from HER2+ EVs and 0.75 for miR-223-3p from CD24+ EVs. LASSO classification selected a panel of four complementary EV miRNA for classifying breast cancer: miR-340-5p (HER2+ EVs), miR-598-3p (CD24+), miR-15b-5p (HER2+), and miR-126-3p (CD24+). CONCLUSIONS: HER2+ and CD24+ EV subpopulations contain complementary biomarkers suitable for validation in larger studies that can accurately detect early-stage breast cancer among women with BI-RADS category 4 breast lesions.

  PublisherOA PDFDOI

• ACR Appropriateness Criteria® Breast Imaging During Pregnancy

  Journal of the American College of Radiology · 2025-11-01 · 2 citations

  article
  PublisherDOI

• Convolutional Neural Network Model Observers Discount Signal-like Anatomical Structures During Search in Virtual Digital Breast Tomosynthesis Phantoms

  arXiv (Cornell University) · 2024-05-23 · 1 citations

  preprintOpen access

  Model observers are computational tools to evaluate and optimize task-based medical image quality. Linear model observers, such as the Channelized Hotelling Observer (CHO), predict human accuracy in detection tasks with a few possible signal locations in clinical phantoms or real anatomic backgrounds. In recent years, Convolutional Neural Networks (CNNs) have been proposed as a new type of model observer. What is not well understood is what CNNs add over the more common linear model observer approaches. We compare the CHO and CNN detection accuracy to the radiologist's accuracy in searching for two types of signals (mass and microcalcification) embedded in 2D/3D breast tomosynthesis phantoms (DBT). We show that the CHO model's accuracy is comparable to the CNN's performance for a location-known-exactly detection task. However, for the search task with 2D/3D DBT phantoms, the CHO's detection accuracy was significantly lower than the CNN accuracy. A comparison to the radiologist's accuracy showed that the CNN but not the CHO could match or exceed the radiologist's accuracy in the 2D microcalcification and 3D mass search conditions. An analysis of the eye position showed that radiologists fixated more often and longer at the locations corresponding to CNN false positives. Most CHO false positives were the phantom's normal anatomy and were not fixated by radiologists. In conclusion, we show that CNNs can be used as an anthropomorphic model observer for the search task for which traditional linear model observers fail due to their inability to discount false positives arising from the anatomical backgrounds.

  PublisherOA PDFDOI

• Abbreviated Breast MRI for Supplemental Screening in Patients With Dense Breasts: Comparison of Baseline Versus Subsequent-Round Examinations

  American Journal of Roentgenology · 2024-05-22 · 5 citations

  article

  BACKGROUND. Abbreviated breast MRI (AB-MRI) achieves a higher cancer detection rate (CDR) than digital breast tomosynthesis when applied for baseline (i.e., first-round) supplemental screening of individuals with dense breasts. Limited literature has evaluated subsequent (i.e., sequential) AB-MRI screening rounds.

  PublisherDOI

• CPI Editor's Choice 2024: Breast Imaging

  2024-02-21

  dataset
  PublisherDOI

Frequent coauthors

• Emily F. Conant

  Hospital of the University of Pennsylvania

  63 shared

• Kevin R. Fox

  36 shared

• Rosemarie Mick

  32 shared

• Shuwen Xu

  Anhui Academy of Medical Sciences

  32 shared

• Angela DeMichele

  University of Pennsylvania

  30 shared

• Paul J. Zhang

  Hospital of the University of Pennsylvania

  29 shared

• Brian J. Czerniecki

  Moffitt Cancer Center

  29 shared

• Elizabeth S. McDonald

  University of Pennsylvania

  27 shared