The objective of the study was to use a deep learning model to differentiate between benign and malignant sentinel lymph nodes (SLNs) in patients with breast cancer compared to radiologists' assessments. Seventy-nine women with breast cancer were enrolled and underwent lymphosonography and contrast-enhanced ultrasound (CEUS) examination after subcutaneous injection of ultrasound contrast agent around their tumor to identify SLNs. Google AutoML was used to develop image classification model. Grayscale and CEUS images acquired during the ultrasound examination were uploaded with a data distribution of 80% for training/20% for testing. The performance metric used was area under precision/recall curve (AuPRC). In addition, 3 radiologists assessed SLNs as normal or abnormal based on a clinical established classification. Two-hundred seventeen SLNs were divided in 2 for model development; model 1 included all SLNs and model 2 had an equal number of benign and malignant SLNs. Validation results model 1 AuPRC 0.84 (grayscale)/0.91 (CEUS) and model 2 AuPRC 0.91 (grayscale)/0.87 (CEUS). The comparison between artificial intelligence (AI) and readers' showed statistical significant differences between all models and ultrasound modes; model 1 grayscale AI versus readers, P = 0.047, and model 1 CEUS AI versus readers, P < 0.001. Model 2 r grayscale AI versus readers, P = 0.032, and model 2 CEUS AI versus readers, P = 0.041. The interreader agreement overall result showed kappa values of 0.20 for grayscale and 0.17 for CEUS. In conclusion, AutoML showed improved diagnostic performance in balance volume datasets. Radiologist performance was not influenced by the dataset's distribution.

After clinical evaluation, especially clinical prediction rules, appropriately ordered venous duplex has become the standard test for evaluating and excluding deep vein thrombosis (DVT). Ultrasound is useful for lower- and upper-extremity veins. Protocols include grayscale, color Doppler, and spectral Doppler. Recommended lower-extremity protocols include central leg and calf veins. Duplex Doppler is widely used to evaluate patients with chronic venous disease, especially with suspected venous reflux. Mapping to identify adequate veins before surgery is another widely used indication for venous ultrasound. Ultrasound for thrombosis can be characterized as normal, acute DVT, superficial thrombosis, or chronic postthrombotic changes in most patients.

Peripheral arterial ultrasound is an important technique to evaluate vascular disease. Protocols include grayscale, color Doppler and spectral Doppler. Arterial duplex is used to evaluate patients with claudication to characterize atherosclerotic disease. Stenoses have characteristic findings which can help profile lesions accurately. Other nonatherosclerotic obstructive processes are also evaluable by duplex Doppler. Traumatic injuries can also be studied in the stable patient. Attention to protocols helps ensure optimal results.

Approximately 15%-25% of breast lymphatic drainage passes through the internal thoracic (internal mammary) lymphatic system, draining the inner quadrants of the breast. This study aimed to use lymphosonography to identify sentinel lymph nodes (SLNs) in the axillary and internal thoracic lymphatic systems in patients with breast cancer. Seventy-nine patients received subcutaneous ultrasound contrast agent injections around the tumor. Lymphosonography was used to identify SLNs. In 14 of the 79 patients (17.7%), the tumor was located in the inner quadrant of the breast. Lymphosonography identified 217 SLNs in 79 patients, averaging 2.7 SLNs per patient. The 217 identified SLNs in the 79 patients were located in the axillary lymphatic system; none were located in the internal thoracic (internal mammary) lymphatic system, although it was expected in two to four patients (i.e., 4-11 SLNs). These results implied that SLNs associated with breast cancer are predominantly located in the axillary lymphatic system.

objective of the work described here was to evaluate the efficacy of lymphosonography in identi-fying sentinel lymph nodes (SLNs) in patients with breast cancer undergoing surgical excision. Of the 86 individ-uals enrolled, 79 completed this institutional review board-approved study. Participants received subcutaneous 1.0-mL injections of ultrasound contrast agent (UCA) around the tumor. An ultrasound scanner with contrast-enhanced ultrasound (CEUS) capabilities was used to identify SLNs. Participants were administered with blue dye and radioactive tracer to guide SLN excision as standard-of-care. Excised SLNs were classified as positive or negative for the presence of blue dye, radioactive tracer and UCA, and sent for pathology. Two hundred fifty-two SLNs were excised; 158 were positive for blue dye, 222 were positive for radioactive tracer and 223 were positive for UCA. Comparison with blue dye revealed accuracies of 96.2% for radioactive tracer and 99.4% for lympho-sonography (p > 0.15). Relative to radioactive tracer, blue dye had an accuracy of 68.5%, and lymphosonography achieved 86.5% (p < 0.0001). Of 252 SLNs excised, 34 were determined to be malignant by pathology; 18 were positive for blue dye (detection rate = 53%), 23 for radioactive tracer (detection rate = 68%) and 34 for UCA (detection rate = 100%) (p < 0.0001). Lymphosonography was similar in accuracy to radioactive tracer and higher in accuracy than blue dye in identifying SLNs. All 34 malignant SLNs were identified by lymphosonogra-phy. (E-mail: flemming.forsberg@jefferson.edu)(c) 2022 World Federation for Ultrasound in Medicine & Biology. All rights reserved.

ObjectivesThis study evaluated the efficacy of lymphosonography in the identification of sentinel lymph nodes (SLNs) in post neoadjuvant chemotherapy patients with breast cancer scheduled to undergo surgical excision.MethodsSeventy‐nine subjects scheduled for breast cancer surgery with SLN excision completed this IRB‐approved study, out of which 18 (23%) underwent neoadjuvant chemotherapy before surgery. Subjects underwent percutaneous Sonazoid (GE Healthcare) injections around the tumor area for a total of 1.0 mL. Lymphosonography was performed using CPS on an S3000 HELX scanner (Siemens Healthineers) with a linear probe. Subjects received blue dye and radioactive tracer as part of their standard of care. Excised SLNs were classified as positive or negative for the presence of blue dye, radioactive tracer and Sonazoid. The results were compared between methods and pathology findings.ResultsSeventy‐two SLNs were surgically excised from 18 subjects, 29 were positive for blue dye, 63 were positive for radioactive tracer and 57 were positive for Sonazoid. Comparison with blue dye showed that both radioactive tracer and lymphosonography achieved an accuracy of 53% (P > .50). Comparison with radioactive tracer showed that blue dye had an accuracy of 53%, while lymphosonography achieved an accuracy of 67% (P < .01). Of the 72 SLNs, 15 were determined malignant by pathology; the detection rate was 47% for blue dye (7/15), 67% for radioactive tracer (10/15) and 100% for lymphosonography (15/15) (P < .001).ConclusionsLymphosonography achieved similar accuracy as radioactive tracer and higher accuracy than blue dye for identifying SLNs. The 15 SLNs positive for malignancy were all identified by lymphosonography.

The 2019 novel coronavirus, known as COVID-19, has greatly affected the way sonographers care for their patients. Sonography can be a useful imaging tool for surveillance and diagnosis of various conditions associated with COVID-19 patients or patients under investigation (PUIs). Currently, there are limited resources and protocols for preventing the transmission of COVID-19 from ultrasound equipment. Our institution has created a detailed protocol for scanning COVID-19 patients or PUIs to address this important issue.

Diagnostic criteria to classify severity of internal carotid artery (ICA) stenosis vary across vascular laboratories. Consensus-based criteria, proposed by the Society of Radiologists in Ultrasound in 2003 (SRUCC), have been broadly implemented but have not been adequately validated. We conducted a multicentered, retrospective correlative imaging study of duplex ultrasound versus catheter angiography for evaluation of severity of ICA stenosis. Velocity data were abstracted from bilateral duplex studies performed between 1/1/2009 and 12/31/2015 and studies were interpreted using SRUCC. Percentage ICA stenosis was determined using North American Symptomatic Carotid Endarterectomy Trial (NASCET) methodology. Receiver operating characteristic analysis evaluated the performance of SRUCC parameters compared with angiography. Of 448 ICA sides (from 224 patients), 299 ICA sides (from 167 patients) were included. Agreement between duplex ultrasound and angiography was moderate (κ = 0.42), with overestimation of degree of stenosis for both moderate (50–69%) and severe (⩾ 70%) ICA lesions. The primary SRUCC parameter for ⩾ 50% ICA stenosis of peak-systolic velocity (PSV) of ⩾ 125 cm/sec did not meet prespecified thresholds for adequate sensitivity, specificity, and accuracy (sensitivity 97.8%, specificity 64.2%, accuracy 74.5%). Test performance was improved by raising the PSV threshold to ⩾ 180 cm/sec (sensitivity 93.3%, specificity 81.6%, accuracy 85.2%) or by adding the additional parameter of ICA/common carotid artery (CCA) PSV ratio ⩾ 2.0 (sensitivity 94.3%, specificity 84.3%, accuracy 87.4%). For ⩾ 70% ICA stenosis, analysis was limited by a low number of cases with angiographically severe disease. Interpretation of carotid duplex examinations using SRUCC resulted in significant overestimation of severity of ICA stenosis when compared with angiography; raising the PSV threshold for ⩾ 50% ICA stenosis to ⩾ 180 cm/sec as a single parameter or requiring the ICA/CCA PSV ratio ⩾ 2.0 in addition to PSV of ⩾ 125 cm/sec for laboratories using the SRUCC is recommended to improve the accuracy of carotid duplex examinations.

The novel coronavirus disease-2019 (COVID-19) illness and deaths, caused by the severe acute respiratory syndrome coronavirus-2, continue to increase. Multiple reports highlight the thromboembolic complications, such as pulmonary embolism (PE), in COVID-19. Imaging plays an essential role in the diagnosis and management of COVID-19 patients with PE. There continues to be a rapid evolution of knowledge related to COVID-19 associated PE. This review summarises the current understanding of prevalence, pathophysiology, role of diagnostic imaging modalities, and management, including catheter-directed therapy for COVID-19 associated PE. It also describes infection control considerations for the radiology department while providing care for patients with COVID-19 associated PE.

chapter in textbook dealing with the diagnosis of stenosis and occlusion of ICA and cerebral arteries, as well as vertebral and basilar arteries

The hepatitis B virus (HBV) is one of the leading causes of hepatocellular carcinoma(HCC) worldwide. In the endemic region, the infection is commonly spread through verticaltransmission in which mother and child possess genetically identical viral genotypes in the settingof similar host genomes. Despite these genetic similarities, clinical outcomes from chronic hepatitisB (CHB) can vary widely, ranging from lifelong asymptomatic infection to terminal HCC. Presentedhere are the longitudinal observations over multiple decades of three family clusters, includingmonozygotic twins with non-discordant HCC, that demonstrate the heterogeneity ofHBV-related outcomes. These findings emphasize the important need to untangle the role of geneticand non-genetic host factors in the development of HBV-related HCC, as well as highlightthe novel research avenues that can clarify the contributions of such factors in HBV-related HCC.

Objective This study evaluated the performance of automated machine-learning to diagnose non-alcoholic fatty liver disease (NAFLD) by ultrasound and compared these findings to radiologist performance. Methods 96 patients with histologic (33) or proton density fat fraction MRI (63) diagnosis of NAFLD and 100 patients without evidence of NAFLD were retrospectively identified. The u201CFatty Liveru201D label included 96 patients with 405 images and the u201CNot Fatty Liveru201D label included 100 patients with 500 images. These 905 images made up a u201CComprehensive Imageu201D group. A u201CRadiology Selected Imageu201D group was then created by selecting only images considered diagnostic by a blinded radiologist, resulting in 649 images. Cloud AutoML Visionbeta (Google LLC, Mountain View, CA) was used for machine learning. The models were evaluated against three blinded radiologists. Results The u201CComprehensive Imageu201D group model demonstrated a sensitivity of 88.6% (73.3u201396.8%) and a specificity of 95.3% (84.2u201399.4%). Radiologist performance on this image group included a sensitivity of 81.0% (74.3u201387.6%) and specificity of 86.0% (72.6u201399.5%). The modelu2019s overall accuracy was 92.3% (84.0u201397.1%), compared with mean individual performance (83.8%, 78.4u201389.1%). The u201CRadiology Selected Imageu201D group model demonstrated a sensitivity of 88.6% (73.3 u2013 96.8%) and specificity of 87.9% (71.8u201396.6%). Mean radiologist sensitivity was 92.4% (86.9u201397.9%) and specificity was 91.9% (83.4u2013100%). The modelu2019s overall accuracy was 88.2% (78.1u201394.8%) which was comparable to the individual radiologist performance (92.2%, 90.1u201394.2%) and consensus performance (95.6%, 87.6u201399.1%). Conclusions An automated machine-learning algorithm may accurately detect NAFLD on ultrasound.

Chapter in US textbook which is the introduction to vascular US with the emphasis on basic applied ultrasound physics, and interpretation of Doppler spectra in various pathologic conditions