Deep learning detection and quantification of pneumothorax in heterogeneous routine chest computed tomography

In this study, the authors proposed a deep learning method for the detection and quantification of pneumothorax in heterogeneous routine clinical data, which may facilitate the automated triage of urgent examinations and enable support in the treatment decision.

Key points

• Pneumothorax is an important pathology to be included in applications that are designed to triage urgent imaging examinations.
• Heterogeneity in routine clinical data may be overcome by utilising deep learning methods.
• Additional automated quantification of pneumothorax volume correlates well with manual volumetric assessment, but is less time-consuming.

Article: Deep learning detection and quantification of pneumothorax in heterogeneous routine chest computed tomography

Authors: Sebastian Röhrich, Thomas Schlegl, Constanze Bardach, Helmut Prosch & Georg Langs

Written by

European Radiology Experimental