Deep Learning Systems for Pneumothorax Detection on Chest Radiographs: A Multicenter External Validation Study.

PURPOSE: To assess the generalizability of a deep learning pneumothorax detection model on datasets from multiple external institutions and examine patient and acquisition factors that might influence performance. MATERIALS AND METHODS: In this retrospective study, a deep learning model was trained for pneumothorax detection by merging two large open-source chest radiograph datasets: ChestX-ray14 and CheXpert. It was then tested on six external datasets from multiple independent institutions (labeled A-F) in a retrospective case-control design (data acquired between 2016 and 2019 from institutions A-E; institution F consisted of data from the MIMIC-CXR dataset). Performance on each dataset was evaluated by using area under the receiver operating characteristic curve (AUC) analysis, sensitivity, specificity, and positive and negative predictive values, with two radiologists in consensus being used as the reference standard. Patient and acquisition factors that influenced performance were analyzed. RESULTS: The AUCs for pneumothorax detection for external institutions A-F were 0.91 (95% CI: 0.88, 0.94), 0.97 (95% CI: 0.94, 0.99), 0.91 (95% CI: 0.85, 0.97), 0.98 (95% CI: 0.96, 1.0), 0.97 (95% CI: 0.95, 0.99), and 0.92 (95% CI: 0.90, 0.95), respectively, compared with the internal test AUC of 0.93 (95% CI: 0.92, 0.93). The model had lower performance for small compared with large pneumothoraces (AUC, 0.88 [95% CI: 0.85, 0.91] vs AUC, 0.96 [95% CI: 0.95, 0.97]; P = .005). Model performance was not different when a chest tube was present or absent on the radiographs (AUC, 0.95 [95% CI: 0.92, 0.97] vs AUC, 0.94 [95% CI: 0.92, 0.05]; P > .99). CONCLUSION: A deep learning model trained with a large volume of data on the task of pneumothorax detection was able to generalize well to multiple external datasets with patient demographics and technical parameters independent of the training data.Keywords: Thorax, Computer Applications-Detection/DiagnosisSee also commentary by Jacobson and Krupinski in this issue.Supplemental material is available for this article.©RSNA, 2021.

Dual-Source Dual-Energy CT in Submandibular Sialolithiasis: Reliability and Radiation Burden.

OBJECTIVE. This study aims to compare the diagnostic accuracy of virtual unenhanced CT images derived from dual-source dual-energy contrast-enhanced CT with that of standard unenhanced CT images for evaluation of sialolithiasis. MATERIALS AND METHODS. All dual-energy CT studies of the neck performed during the preceding 5 years were reviewed for submandibular gland calculi. Only patients who had unenhanced CT and contrast-enhanced CT performed as part of the same evaluation were included in this study. This review yielded 30 patients. Virtual unenhanced CT images were derived from the dual-energy dataset and compared with the true unenhanced CT images by two separate radiologists who assessed the total number of calculi encountered, their location, the largest dimension, and the attenuation of the calculi. The radiation burden incurred for true unenhanced CT and virtual unenhanced CT and the total radiation burden were calculated. RESULTS. Our analysis revealed that measurements of stone size showed good interobserver agreement. The mean stone size was 7.9 ± 5.9 (SD) mm on virtual unenhanced CT and 8.4 ± 5.9 mm on true unenhanced CT (range, 2.0-31.8 mm); the difference was statistically insignificant. The mean stone attenuation differed considerably (p < 0.01) between the virtual and true unenhanced CT images (494.8 ± 187.5 HU and 924.4 ± 374.9 HU, respectively), but correlated well. The mean radiation dose for a dual-phase IV contrast-enhanced CT study was 23.13 mGy (volume CT dose index). The mean dose was 10.93 mGy for the true unenhanced CT phase, thereby suggesting a 47.25% reduction in administered radiation dose when a single-phase contrast-enhanced CT study with virtual unenhanced image reconstructions is performed. CONCLUSION. Virtual unenhanced CT images derived from dual-source dual-energy CT scans of the neck provide accurate assessment of sialolithiasis akin to that provided by conventional CT protocols but at only a fraction of the radiation dosage.

Imaging Sinonasal disease with MRI: Providing insight over and above CT.

This article illustrates and discusses the applications and value of magnetic resonance imaging (MRI) in the evaluation of sinonasal disease. There are several clinical scenarios where MRI can add value over conventional computed tomography (CT) evaluation of the sinonasal spaces. Specifically, MRI can provide insight through better depiction of the anatomy of certain sinonasal sub-sites including the olfactory structures. It can aid in evaluating anosmia, sinusitis (fungal sinusitis and complications), benign and malignant lesions, CSF leaks and pathology extending into sinonasal spaces.

Tension pneumocephalus.

Tension pneumocephalus is a rare but treatable neurosurgical emergency. Prompt and accurate diagnosis of tension pneumocephalus requires a high index of clinical suspicion corroborated by imaging. Herein, we describe a case of extensive tension pneumocephalus in a patient who had undergone transsphenoidal surgery and repair of the sellar floor, with subsequent successful decompression. This case report discusses the pertinent imaging features of tension pneumocephalus and its management.