AJR: CT may accelerate and improve diagnosis of gunfire trauma
Radiologists in Iraq and Afghanistan have made extensive use of CT to quickly diagnose and triage soldiers injured by gunfire and blasts. “Ballistic injuries, however, are not unique to the military and are not rare in civilian settings. Penetrating trauma has gained importance in civilian care as a result of increased worldwide terrorist activity. ... Radiologists, therefore, play a central role in the diagnosis, triage and management of blast and ballistic trauma,” wrote Les R. Folio, DO, MPH, from the department of radiology at the National Institutes of Health in Bethesda, Md., and colleagues.
Folio and co-authors investigated the utility of wound path trajectory analysis on multi-detector CT (MDCT). Three radiologists reviewed 25 deep-penetrating trajectories from 19 casualties in Iraq in order to compare the qualitative and quantitative (angular) trajectories of projectiles.
Eleven head and 14 body trajectories were included in the study. Reviewers interpreted axial CT images qualitatively to determine the directional path of the bullet, such as anteroposterior vs. posteroanterior or right-to-left vs left-to-right. Trajectories were also assessed quantitatively, using a 3D Cartesion positioning system, as vectors across the axial image, with traditional x- and y-axes, plus a z-axis corresponding to the table’s position, or sagittal plane.
All observers agreed on all three qualitative descriptors for eight of 11 head and nine of 14 body trajectories, making for 68 percent total agreement on qualitative bullet direction.
Observer calculations agreed with all three qualitative directions in six of 11 head trajectories and nine of 14 body trajectories. Overall, 60 of 75 individual calculated descriptors were identical with the visually assessed descriptors. Kappa statistics showed good agreement (0.65 to 0.79) and were larger for visually assessed than for calculated descriptors.
The researchers considered quantitative calculations of trajectory angles to be in agreement if they measured within 20 degrees of one another. Agreement was independently reached in 16 of 22 parasagittal and paraaxial intracranial trajectory angles and 19 of 28 pairs of body trajectory angles.
For angles between trajectories, 46 of 75 pairs of trajectory vectors derived from the 25 intracranial and body trajectories from the three pairs of reviewer calculations fell within 20 degrees.
“Most calculated trajectory angles (paraaxial and parasagittal angles) and most of the angles between calculated trajectories agreed within 20 degrees across observers. This indicates reasonable accuracy in determining interobserver agreement in the context of penetrating trauma resulting from large irregular projectiles as encountered in our study population,” Folio and colleagues wrote.
The authors argued that, despite the higher volume of penetrating trauma in military casualties, their findings would be generalizable to the civilian population, following further research. Moreover, Folio and colleagues argued that their CT-based Cartesion positioning system method of trajectory analysis could help identify the location of snipers or the origins of missiles by providing information on the direction and angle of bullet entrance.
“Because mathematic and visual determination of directional descriptors of trajectories show similar results, our findings support the feasibility of automated trajectory analysis with minimal input from an observer requiring only basic training,” the authors added.
The application of their findings to non-axial images, larger patient groups and non-linear bullet trajectories represent important obstacles to attaining more accurate semi-automated bullet trajectory calculations, Folio et al stated.
“Surgical exploration and autopsies are often claimed as the reference standard for the evaluation of pathologic findings. However, we believe that especially in penetrating blast and ballistic trauma MDCT with image postprocessing is superior to invasive procedures because it enables depiction of anatomy and locations of all the particles in vivo without destruction or distortion of relevant structures,” Folio and colleagues wrote.
“Therefore, in our study determination of angles between linear estimates of trajectories identified on MDCT provided a reliable summary measure of interobserver agreement of calculated wound path estimates in all planes.”