Carpal pseudoerosions: a plain X-ray interpretation pitfall

To examine in detail images of pseudoerosion of the wrist and hand on plain radiographs. The study was conducted with 28 cadaver wrists. During a single imaging session three techniques-plain radiography, tomosynthesis, and computed tomography-were used to visualize the wrist and hand specimens. For each technique, 20 radio-ulno-carpo-metacarpal sites known to present bone erosions in rheumatoid arthritis were analyzed by two radiologists using a standard system to score the cortical bone: normal, pseudoerosion, true erosion, or other pathology. Cohen's concordance analysis was performed to determine inter-observer and intra-observer (for the senior radiologist) agreement by site and by technique. Serial sections of two cadaver specimens were examined to determine the anatomical correlation of the pseudoerosions. On the plain radiographs, the radiologists scored many images as pseudoerosion (7.3 %), particularly in the distal ulnar portion of the capitate, the distal radial portion of the hamate, the proximal ulnar portion of the base of the third metacarpal, the proximal radial portion of the base of the fourth metacarpal, the distal ulnar portion of the hamate, and the proximal portion of the base of the fifth metacarpal. The computed tomography scan revealed that none of these doubtful images corresponded to true erosions. The anatomical correlation study showed that these images could probably be attributed to ligament insertions, thinner lamina, and enhanced cortical bone transparency. Knowledge of the anatomical carpal localizations where pseudoerosions commonly occur is a necessary prerequisite for analysis of plain radiographs performed to diagnose or monitor rheumatoid arthritis.