An abnormal collection of air in the thorax is one of the most common life-threatening events that occurs in the intensive care unit. Patient management differs depending on the location of the air collection; therefore, detecting abnormal air collection and identifying its exact location on supine chest radiographs is essential for early treatment and positive patient outcomes. Thoracic abnormal air collects in multiple thoracic spaces, including the pleural cavity, chest wall, mediastinum, pericardium, and lung. Pneumothorax in the supine position shows different radiographic findings depending on the location. Many conditions, such as skin folds, interlobar fissure, bullae in the apices, and air collection in the intrathoracic extrapleural space, mimic pneumothorax on radiographs. Additionally, pneumopericardium may resemble pneumomediastinum and needs to be differentiated. Further, some conditions such as inferior pulmonary ligament air collection versus a pneumatocele or pneumothorax in the posteromedial space require a differential diagnosis based on radiographs. Computed tomography (CT) is required to localize the air and delineate potential etiologies when a diagnosis by radiography is difficult. The purposes of this article are to review the anatomy of the potential spaces in the chest where abnormal air can collect, explain characteristic radiographic findings of the abnormal air collection in supine patients with illustrations and correlated CT images, and describe the distinguishing features of conditions that require a differential diagnosis. Since management differs based on the location of the air collection, radiologists should try to accurately detect and identify the location of air collection on supine radiographs.

Adenocarcinoma currently accounts for 10–25% of all uterine cervical carcinomas and has a variety of histopathological subtypes. Among them, mucinous carcinoma gastric type is not associated with high-risk human papillomavirus (HPV) infection and a poor prognosis, while villoglandular carcinoma has an association with high-risk HPV infection and a good prognosis. They show relatively characteristic imaging findings which can be suggested by magnetic resonance imaging (MRI), though the former is sometimes difficult to be distinguished from lobular endocervical glandular hyperplasia. Various kinds of other tumors including squamous cell carcinoma should be also differentiated on MRI, while it is currently difficult to distinguish them on MRI, and HPV screening and pathological confirmation are usually necessary for definite diagnosis and further patient management.

No abstract

Background: This study aimed to compare deep learning with radiologists’ assessments for diagnosing ovarian carcinoma using MRI. Methods: This retrospective study included 194 patients with pathologically confirmed ovarian carcinomas or borderline tumors and 271 patients with non-malignant lesions who underwent MRI between January 2015 and December 2020. T2WI, DWI, ADC map, and fat-saturated contrast-enhanced T1WI were used for the analysis. A deep learning model based on a convolutional neural network (CNN) was trained using 1798 images from 146 patients with malignant tumors and 1865 images from 219 patients with non-malignant lesions for each sequence, and we tested with 48 and 52 images of patients with malignant and non-malignant lesions, respectively. The sensitivity, specificity, accuracy, and AUC were compared between the CNN and interpretations of three experienced radiologists. Results: The CNN of each sequence had a sensitivity of 0.77–0.85, specificity of 0.77–0.92, accuracy of 0.81–0.87, and an AUC of 0.83–0.89, and it achieved a diagnostic performance equivalent to the radiologists. The CNN showed the highest diagnostic performance on the ADC map among all sequences (specificity = 0.85; sensitivity = 0.77; accuracy = 0.81; AUC = 0.89). Conclusion: The CNNs provided a diagnostic performance that was non-inferior to the radiologists for diagnosing ovarian carcinomas on MRI.

No abstract