Susan P. Weinstein

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.

Despite the benefits of breast MRI in the detection of cancer, false-negative results may occur, but radiologists can avoid these errors by understanding the causes of false-negative findings at br... Breast MRI is the most sensitive modality for the detection of breast cancer. However, false-negative cases may occur, in which the cancer is not visualized at MRI and is instead diagnosed with ano...

PURPOSE: Although mammography is the standard of care for breast cancer screening, dense breast tissue decreases mammographic sensitivity. We report the prevalent cancer detection rate (CDR) from the first clinical implementation of abbreviated breast magnetic resonance imaging (AB-MR) as a supplemental screening test in women with dense breasts. METHODS: The study was approved by the institutional review board and is Health Insurance Portability and Accountability Act complaint. This retrospective review includes women who were imaged between January 1, 2016 and February 28, 2019. On a 1.5 Tesla magnet, the imaging protocol consisted of three sequences: Short-TI Inversion Recovery (STIR), precontrast, and postcontrast. A subtraction sequence and a maximum intensity projection were generated. We report the patient-level CDR and the positive predictive value of AB-MR examinations after negative/benign digital breast tomosynthesis (DBT). RESULTS: Out of 511 prevalent rounds of AB-MR examinations, 36 women were excluded. The remaining 475 asymptomatic women with dense breasts had negative/benign DBT examinations before the AB-MR. There were 420 of 475 (88.4%) benign/negative examinations, 13 of 475 (2.7%) follow-up recommendations, and 42 biopsy recommendations. Thirty-nine biopsies were completed, resulting in 12/39 (30.8%) malignancies in 12 women: seven invasive carcinomas and five ductal carcinoma in situ. One additional patient was diagnosed with invasive ductal carcinoma at the time of 6-month follow-up. The CDR was 27.4 per 1,000 (13 of 475; 95% CI, 16.1 to 46.3). The size of invasive carcinomas ranged from 0.6-1.0 cm (mean, 0.5 cm). Of the seven women who underwent surgical evaluation of the axilla, zero of seven patients had positive nodes. There were no interval cancers at 1-year follow-up. CONCLUSION: Preliminary results from clinical implementation of screening AB-MR resulted in a CDR of 27.4/1,000 at the patient level after DBT in women with dense breasts. Additional evaluation is warranted.

The sensitivity of digital breast tomosynthesis (DBT) was higher than that of digital mammography (DM) in each of the first 5 years after implementation of DBT. DBT also helped detect a higher prop... Background Limited data exist beyond prevalence rounds of digital breast tomosynthesis (DBT) screening. Purpose To compare DBT outcomes over multiple years and rounds to outcomes of digital mammogr...

This article highlights the current concepts of digital breast tomosynthesis imaging, including synthetic mammography, that are relevant to contemporary breast imaging; the quasi three-dimensional ... Digital breast tomosynthesis (DBT) is emerging as the standard of care for breast imaging based on improvements in both screening and diagnostic imaging outcomes. The additional information obtaine...

For cancers visible at only one mammographic view, they are more often seen at the craniocaudal view than the mediolateral oblique view for both digital mammography and digital breast tomosynthesis. Background Digital breast tomosynthesis (DBT) has been shown to improve screening outcomes compared with digital mammography (DM) alone. However, little is known about differences in breast cancer ...

This study introduces a fully automated method for the computation of quantitative volumetric breast density from digital breast tomosynthesis (DBT) images and demonstrates a comparison with volume-based measures from full-field digital mammography, DBT, and MR imaging. PurposeTo assess a fully automated method for volumetric breast density (VBD) estimation in digital breast tomosynthesis (DBT) and to compare the findings with those of full-field digital mammography (FFDM) and magnetic resonance (MR) imaging.Materials and MethodsBilateral DBT images, FFDM images, and sagittal breast MR images were retrospectively collected from 68 women who underwent breast cancer screening from October 2011 to September 2012 with institutional review board–approved, HIPAA-compliant protocols. A fully automated computer algorithm was developed for quantitative estimation of VBD from DBT images. FFDM images were processed with U.S. Food and Drug Administration–cleared software, and the MR images were processed with a previously validated automated algorithm to obtain corresponding VBD estimates. Pearson correlation and analysis of variance with Tukey-Kramer post hoc correction were used to compare the multimodality VBD estimates.ResultsEstimates of VBD from DBT were significantly correlat...