Dark-Field Chest Radiography for Pneumothorax Detection: A Prospective Study

Purpose To evaluate the potential of dark-field chest radiography to improve the detection and assessment of pneumothorax. Materials and Methods This prospective study was conducted between March 2022 and September 2023. Participants with or without pneumothorax underwent chest radiography using a prototype system that acquires attenuation-based and dark-field images simultaneously. Five blinded readers independently assessed attenuation-only radiographs and then, after a 4-week interval, assessed dark-field overlay images. Sensitivity, specificity, accuracy, reading time, and diagnostic confidence for pneumothorax detection were compared between the two approaches using appropriate statistical tests. Results The sample included 100 participants (median age, 62 years; IQR, 52-71 years; 61 male participants), including 36 individuals with clinically confirmed pneumothorax and 64 healthy controls. A nonsignificant increase in sensitivity for pneumothorax detection was observed with dark-field overlays (87.4% [95% CI: 67.5, 95.8]) compared with attenuation-based radiographs (84.2% [95% CI: 73.2, 91.2], P = .61), whereas specificity was nearly identical (97.6% [95% CI: 94.0, 99.1] vs 97.5 [95% CI: 95.7, 98.6], P = .89). The median reading time decreased from 30.8 seconds to 10.3 seconds (P < .001), and diagnostic confidence improved across all readers (median score: 3 vs 4; P < .001). Conclusion The addition of dark-field radiography to conventional chest radiography improves pneumothorax detection, substantially reduces reading time, and enhances diagnostic confidence without loss of sensitivity and specificity. Keywords: Pneumothorax, Radiography, Dark-Field © RSNA, 2025.