Improved Detection of Foreign Bodies on Radiographs Using X-ray Dark-Field and Phase-Contrast Imaging

Invest Radiol. 2018 Jun;53(6):352-356. doi: 10.1097/RLI.0000000000000450.

Abstract

Purpose: The aim of this study was to investigate whether the detection of foreign bodies can be improved using dark-field and phase-contrast radiography compared with conventional (transmission) radiographs.

Materials and methods: Experiments were performed using ex vivo pig paws, which were prepared with differently sized foreign bodies of metal, wood, and glass (n = 10 each). Paws without foreign bodies served as controls (n = 30). All images were acquired using an experimental grating-based large object radiography system. Five blinded readers (second- to fourth-year radiology residents) were asked to assess the presence or absence of any foreign body. Sensitivity and specificity for the detection of metal, wood, glass, and any foreign body were calculated and compared using McNemar test and generalized linear mixed models.

Results: Sensitivity for the detection of metal foreign bodies was 100% for all readers and image combinations. The sensitivity for the detection of wooden foreign bodies increased from 2% for transmission images to 78% when dark-field images were added (P < 0.0001). For glass foreign bodies, sensitivity increased from 84% for transmission images to 96% when adding phase-contrast images (P = 0.041). Sensitivity for the detection of any foreign body was 91% when transmission, dark-field, and phase-contrast images were viewed simultaneously, compared with 62% for transmission images alone (P < 0.0001). Specificity was 99% to 100% across all readers and radiography modalities.

Conclusions: Adding dark-field images substantially improves the detection of wooden foreign bodies compared with the analysis of conventional (transmission) radiographs alone. Detection of glass foreign bodies was moderately improved when adding phase-contrast images.

MeSH terms

  • Animals
  • Foreign Bodies / diagnostic imaging*
  • Image Processing, Computer-Assisted / methods*
  • Models, Animal
  • Radiography / methods*
  • Sensitivity and Specificity
  • Swine