Ultrasound versus radiography in the detection of soft-tissue foreign bodies

To determine the usefulness of ultrasound and radiography in detecting foreign bodies in soft-tissue models closely duplicating puncture-wound trauma and hand anatomy.

In this randomized, blinded descriptive study, two radiologists independently evaluated 120 chicken thighs for foreign bodies with the use of standard two-view radiography and 7.5-MHz transducer ultrasonography. All chicken thighs were manipulated with hemostats to ensure uniform tissue damage. In 60 thighs, one foreign body had been inserted (10 each: gravel, metal, glass, cactus spine, wood, and plastic).

The sensitivity of ultrasound in detecting gravel was 40%, that for metal was 45%, that for glass was 50%, that for cactus spine was 30%, that for wood was 50%, and that for plastic was 40%. The overall sensitivity, specificity, and false-negative and false-positive rates for ultrasound were 43%, 70%, 50%, and 30%, respectively. No individual foreign body had an ultrasound detection rate of 50%. Radiography detected foreign bodies generally considered radiopaque (gravel, glass, metal) 98% of the time, but it never detected bodies considered radiolucent (wood, plastic, cactus spine). The false-negative and false-positive rates for radiography were 50% and 1.6%, respectively.

Ultrasound detection of foreign bodies by skilled operators in this animal model revealed poor sensitivity and specificity. Radiographic detection was highly sensitive for foreign bodies considered radiopaque. Our data suggest that ultrasound should not be relied on to rule out the possibility of a retained foreign body in the distal extremities.