Cartesian positioning system for localization of blast and ballistic fragments: a phantom-based pilot study

Our purpose was to demonstrate the consistency of radiologists' three-dimensional measurements of simulated blast fragment locations in vitro in an effort to objectively localize retained fragments and wound paths. We designed a phantom consisting of 10 nail heads (simulating blast fragments) glued to wooden pegs that were randomly situated at distances from a reference point within a plastic tub. The x, y, and z coordinates of simulated fragments were recorded in Cartesian 3-space relative to the reference point. Computed tomography images of the phantom were acquired. Differences in x, y, and z positions as determined by three observers were summed for each fragment. Agreement between recordings of coordinates across readers was assessed using the intraclass correlation coefficient. Summed differences in coordinate positions as determined by readers ranged between 0.00 and 1.204 cm (mean: 0.732 cm). Across readers, the intraclass correlation coefficient for each dimension was >0.99. We found excellent agreement among readers with minimal discrepancy of measured locations of simulated fragments. Our results provide a foundation for trajectory analysis necessary to lead to automated organ damage reporting for immediate assessment in the emergency department and for forensic investigation and long-term epidemiological analysis.