RESUMO
We describe the curation, annotation methodology, and characteristics of the dataset used in an artificial intelligence challenge for detection and localization of COVID-19 on chest radiographs. The chest radiographs were annotated by an international group of radiologists into four mutually exclusive categories, including "typical," "indeterminate," and "atypical appearance" for COVID-19, or "negative for pneumonia," adapted from previously published guidelines, and bounding boxes were placed on airspace opacities. This dataset and respective annotations are available to researchers for academic and noncommercial use.
Assuntos
COVID-19 , Humanos , Inteligência Artificial , Radiografia , Aprendizado de Máquina , Radiologistas , Radiografia Torácica/métodosRESUMO
INTRODUCTION: Voiding cystourethrogram (VCUG) with fluoroscopy remains the gold standard for detection and evaluation of vesicoureteral reflux (VUR) among children. However, the ionizing radiation exposure remains a concern for this diagnostic modality. Recent studies have proposed using contrast-enhanced ultrasound as an alternative option for VUR screening and follow-up in children. The aim of the study was to review the literature of comparative studies that assessed the diagnostic accuracy of contrast-enhanced ultrasound compared with VCUG. METHODOLOGY: A systematic literature search was performed on electronic medical literature databases in July 2017. Literature identification, screening, and assessment of eligibility were performed by five reviewers with a pediatric radiologist. Literature was summarized for the study population, contrast used, and ultrasound mode as well as the timing of comparative reference study being performed. The studies were clustered according to the kind of contrast used. Reported diagnostic accuracy was extracted from individual studies and summarized across the included studies using descriptive statistics of median and interquartile range (IQR). RESULT: A total of 45 comparative studies were identified as eligible for the summary of the literature. Two generations of ultrasound contrast were identified in the available studies (first generation, Levovist and second generation, SonoVue). For the ultrasound studies using the first-generation contrast, the median sensitivity, regardless of the ultrasound mode, was 90.25 (IQR 83.25-97), and the median specificity was 93 (IQR 91.3-95.25). Among studies using the second-generation contrast, the median sensitivity was 86.26 (IQR 81.13-97), and the median specificity was 90.99 (IQR 84-98). No serious adverse events were reported in any of the studies. CONCLUSION: Overall, this review highlights the application of contrast-enhanced ultrasound for its advantage of no exposure to ionizing radiation and diagnostic accuracy relatively comparable to VCUG in the evaluation of VUR. In addition to the functional evaluation of the VUR, it also provides an anatomic evaluation of the kidneys and bladder with ultrasound imaging. However, one should also note that this alternate procedure is highly operator dependent where diagnostic accuracy is excellent when the expertise is available.