Objectives-To evaluate the diagnostic performance of ultrasound (US)-guided fine-needle aspiration with optional core needle biopsy of head and neck lymph nodes and masses, with attention to differences between biopsy of treated squamous cell carcinoma (SCC) and biopsy of other lesions.Methods-Institutional Review Board approval was obtained, and the need for consent was waived for this retrospective study. All 861 US-guided biopsies of head and neck lymph nodes and masses performed between March 1, 2012, and May 16, 2016, were reviewed.Results-Of the 861 biopsies, 53 targeted SCC with residual masses after treatment. The biopsy procedures yielded benign or malignant pathologic results in 71.7% (38 of 53) of treated SCC and 90.7% (733 of 808) of all other lesions (P < .001). A reference standard based on subsequent pathologic results or clinical and imaging follow-up was established in 68.4% of procedures. In cases with benign or malignant biopsy results and a subsequent reference standard, the sensitivity values for malignancy were 87.5% (95% confidence interval, 64.0%-96.5%) in treated SCC and 98.3% (95% confidence interval, 96.0%-99.3%) in all other cases (P = .047), and the specificity values were 63.6% (95% confidence interval, 35.4%-84.8%) in treated SCC and 99.5% (95% confidence interval, 97.3%-99.9%) in all other cases (P < .001). There were no major complications related to the biopsy procedures.Conclusions-Excluding treated SCC, US-guided fine-needle aspiration with optional core needle biopsy of head and neck lymph nodes and masses has excellent diagnostic performance. Needle biopsy of head and neck SCC with a residual mass after therapy has a high rate of nondiagnostic samples, suboptimal sensitivity, and poor specificity.

No abstract

performed using PyTorch and 2 NVIDIA K80 GPUs. Receiver operating characteristic (ROC) curves with area under the curve (AUC) and standard diagnostic measures (e.g., sensitivity, specificity, accuracy) were used to evaluate the DCNN's performance. Results: The DCNN achieved AUC of 0.96 for detection of IVC filter. At optimal diagnostic thresholds, the DCNN achieved accuracy of 90.9% and sensitivity and specificity of 88.2% and 93.7%, respectively. Conclusions: A DCNN trained on a small set of images from similar but different imaging modalities (radiographs and fluoroscopic images) are able to automatically detect IVC filters with reasonable diagnostic performance. Our study demonstrates proof-of-concept of deep learning towards this task, which will likely be improved with the use of larger datasets as well as extending towards other types of Interventional devices and hardware.

options at their disposal. Targeted case presentations given to preclinical and clerkship-level medical students interested in primary care specialties were effective at increasing the understanding of the role of IR in their specialty of interest. Student-led presentations using interest groups as a networking platform is an effective method for forming first impressions and exposing future doctors to the role of interventional radiology in their practice.

No abstract