Radiological Profile of 18,430 Vascular Anomalies: Incidence and Demographic Distribution in an Adult Population.

BACKGROUND: Most studies on focusing on the prevalence of vascular anomalies are either aimed to determine the individual occurrence of a specific type among known bearers of abnormalities or propose an estimation of prevalence for the general population by extrapolating from the paediatric population. In this scenario, we intended to assess the profile of vascular anomalies in a group of patients subjected to imaging studies, throughout a long period of time, to evaluate the frequency of abnormal findings in a consecutive, nonselected population. METHODS: This is a retrospective review of 996,569 computed tomography and magnetic resonance studies between 2009 and 2019. Findings were grouped as vascular tumours (hemangiomas; vascular tumours), cavernomas, and vascular malformations. Positive findings were evaluated with regard to patients' demographic characteristics and anatomic distribution and the subset of vascular malformations was also evaluated with regard to size, classification in accordance to flow pattern, and Hamburg Classification. RESULTS: Eighteen thousand four hundred thirty positive examinations were evaluated (mean age, 55.82 ± 15.43 years; 8,188 men). Vascular anomalies were present in 18.49 per 1,000 examinations (17.41 hemangiomas; 0.69 cavernomas and 0.39 vascular malformations per 1,000 examinations). Hemangiomas and cavernomas were homogeneous in anatomic location between sexes throughout the age groups. Complex malformations were heterogeneous in anatomic distribution between the sexes in each age group, with intracranial findings decreasing for female patients in older groups. CONCLUSIONS: Vascular anomalies were found in 18.49 per 1,000 examinations. Hemangiomas and cavernomas were homogenously distributed, whereas complex malformations displayed a heterogeneous anatomic distribution pattern between sexes in each age group.

BRAX, Brazilian labeled chest x-ray dataset.

Chest radiographs allow for the meticulous examination of a patient's chest but demands specialized training for proper interpretation. Automated analysis of medical imaging has become increasingly accessible with the advent of machine learning (ML) algorithms. Large labeled datasets are key elements for training and validation of these ML solutions. In this paper we describe the Brazilian labeled chest x-ray dataset, BRAX: an automatically labeled dataset designed to assist researchers in the validation of ML models. The dataset contains 24,959 chest radiography studies from patients presenting to a large general Brazilian hospital. A total of 40,967 images are available in the BRAX dataset. All images have been verified by trained radiologists and de-identified to protect patient privacy. Fourteen labels were derived from free-text radiology reports written in Brazilian Portuguese using Natural Language Processing.