Faculty Bibliography

3D imaging enables accurate diagnosis by providing spatial information about organ anatomy. However, using 3D images to train AI models is computationally challenging because they consist of 10x or 100x more pixels than their 2D counterparts. To be trained with high-resolution 3D images, convolutional neural networks resort to downsampling them or projecting them to 2D. We propose an effective alternative, a neural network that enables efficient classification of full-resolution 3D medical images. Compared to off-the-shelf convolutional neural networks, our network, 3D Globally-Aware Multiple Instance Classifier (3D-GMIC), uses 77.98%-90.05% less GPU memory and 91.23%-96.02% less computation. While it is trained only with image-level labels, without segmentation labels, it explains its predictions by providing pixel-level saliency maps. On a dataset collected at NYU Langone Health, including 85,526 patients with full-field 2D mammography (FFDM), synthetic 2D mammography, and 3D mammography, 3D-GMIC achieves an AUC of 0.831 (95% CI: 0.769-0.887) in classifying breasts with malignant findings using 3D mammography. This is comparable to the performance of GMIC on FFDM (0.816, 95% CI: 0.737-0.878) and synthetic 2D (0.826, 95% CI: 0.754-0.884), which demonstrates that 3D-GMIC successfully classified large 3D images despite focusing computation on a smaller percentage of its input compared to GMIC. Therefore, 3D-GMIC identifies and utilizes extremely small regions of interest from 3D images consisting of hundreds of millions of pixels, dramatically reducing associated computational challenges. 3D-GMIC generalizes well to BCS-DBT, an external dataset from Duke University Hospital, achieving an AUC of 0.848 (95% CI: 0.798-0.896).

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has a high sensitivity in detecting breast cancer but often leads to unnecessary biopsies and patient workup. We used a deep learning (DL) system to improve the overall accuracy of breast cancer diagnosis and personalize management of patients undergoing DCE-MRI. On the internal test set (n = 3936 exams), our system achieved an area under the receiver operating characteristic curve (AUROC) of 0.92 (95% CI: 0.92 to 0.93). In a retrospective reader study, there was no statistically significant difference (P = 0.19) between five board-certified breast radiologists and the DL system (mean ΔAUROC, +0.04 in favor of the DL system). Radiologists' performance improved when their predictions were averaged with DL's predictions [mean ΔAUPRC (area under the precision-recall curve), +0.07]. We demonstrated the generalizability of the DL system using multiple datasets from Poland and the United States. An additional reader study on a Polish dataset showed that the DL system was as robust to distribution shift as radiologists. In subgroup analysis, we observed consistent results across different cancer subtypes and patient demographics. Using decision curve analysis, we showed that the DL system can reduce unnecessary biopsies in the range of clinically relevant risk thresholds. This would lead to avoiding biopsies yielding benign results in up to 20% of all patients with BI-RADS category 4 lesions. Last, we performed an error analysis, investigating situations where DL predictions were mostly incorrect. This exploratory work creates a foundation for deployment and prospective analysis of DL-based models for breast MRI.

Abbreviated breast MRI is an emerging technique that is being incorporated into clinical practice for breast cancer imaging and screening. Conventional breast MRI includes barriers such as high examination cost and lengthy examination times which make its use in the screening setting challenging. Abbreviated MRI aims to address these pitfalls by reducing overall examination time and increasing accessibility to MRI while preserving diagnostic accuracy. Sequences selected for abbreviated MRI protocols allow for preserved accuracy in breast cancer detection and characterization. Novel techniques such as ultrafast imaging are being used to provide kinetic information from early post-contrast imaging.

Internal mammary lymph nodes constitute a major lymphatic chain draining the breast and a route of spread for breast cancer metastases. Both physiologic and metastatic internal mammary lymph nodes enhance on breast magnetic resonance imaging, and the clinical significance of their prevalence, size, and morphology when visualized in a patient with breast cancer remains unknown. We studied the characteristics of internal mammary lymph nodes visualized on breast MRI studies before and after neo-adjuvant therapy in twenty-three patients with newly diagnosed breast cancer. A measured decrease in internal mammary lymph node size on post-neo-adjuvant therapy MRI indicated metastatic involvement. Determining suspicious features of internal mammary nodes on initial diagnostic MRI can aid radiologists in reporting probable IMLN metastases and may alter the course of care for patients with breast cancer. This study concludes that metastatic internal mammary lymph nodes should be considered when more than two ipsilateral internal mammary lymph nodes measuring 6 mm or greater are seen on diagnostic MRI in a patient with newly diagnosed breast cancer.

Cavities occasionally are encountered on thoracic images. Their differential diagnosis is large and includes, among others, various infections, autoimmune conditions, and primary and metastatic malignancies. We offer an algorithmic approach to their evaluation by initially excluding mimics of cavities and then broadly classifying them according to the duration of clinical symptoms and radiographic abnormalities. An acute or subacute process (< 12 weeks) suggests common bacterial and uncommon nocardial and fungal causes of pulmonary abscesses, necrotizing pneumonias, and septic emboli. A chronic process (≥ 12 weeks) suggests mycobacterial, fungal, viral, or parasitic infections; malignancy (primary lung cancer or metastases); or autoimmune disorders (rheumatoid arthritis and granulomatosis with polyangiitis). Although a number of radiographic features can suggest a diagnosis, their lack of specificity requires that imaging findings be combined with the clinical context to make a confident diagnosis.

Assessment of clinical outcomes such as 30-day mortality following coronary revascularization procedures has historically been used to spur quality improvement programs. Public reporting of risk-adjusted outcomes is already mandated in several states, and proposals to further expand public reporting have been put forward as a means of increasing transparency and potentially incentivizing high quality care. However, for public reporting of outcomes to be considered a useful surrogate of procedural quality of care, several prerequisites must be met. First, the reporting measure must be truly representative of the quality of the procedure itself, rather than be dominated by other underlying factors, such as the overall level of illness of a patient. Second, to foster comparisons among physicians and institutions, the metric requires accurate ascertainment of and adjustment for differences in patient risk profiles. This is particularly relevant for high-risk clinical patient scenarios. Finally, the potential deleterious consequences of public reporting of a quality metric should be considered prior to expanding the use of public reporting more broadly. In this viewpoint, the authors review in particular the characterization of high-risk patients currently treated by percutaneous coronary interventional procedures, assessing the adequacy of clinical risk models used in this population. They then expand upon the limitations of 30-day mortality as a quality metric for percutaneous coronary intervention, addressing the strengths and limitations of this metric, as well as offering suggestions to enhance its future use in public reporting.