RESUMO
PURPOSE: Forensic investigations could benefit from non-invasive in situ characterization of bullets. Therefore, the use of CT imaging was explored for the analysis of bullet geometry and composition. Bullet visualization with CT is challenging as the metal constituents suffer from excessive X-ray attenuation due to their high atomic number, density, and geometry. METHODS: A metal reference phantom was developed containing small discs of various common metals (aluminum, iron, stainless steel, copper, brass, tungsten, and lead). CT images were acquired with 70-150 kVp and 200-400 mAs and were reconstructed using an extended Hounsfield unit (HU) scale (- 10,240 to + 30,710). For each material, the mean CT number (HU) was measured to construct a metal database. Different bullets (n = 11) were scanned in a soft tissue-mimicking phantom. Bullet size and shape were measured, and composition was evaluated by comparison with the metal database. Also, the effect of bullet orientation within the CT scanner was evaluated. RESULTS: In the reference phantom, metals were classified into three groups according to their atomic number (Z): low (Z ≤ 13; HU < 3000), medium (Z = 25-30; HU = 13,000-20,000), and high (Z ≥ 74; HU > 30,000). External bullet contours could be accurately delineated. Internal interfaces between jacket and core could not be identified. Cross-sectional spatial profile plots of the CT number along bullets' short axis revealed beam hardening and photon starvation effects that depended on bullet size, shape, and orientation within the CT scanner. Therefore, the CT numbers of bullets were unreliable and could not be used for material characterization by comparison with the reference phantom. CONCLUSION: CT-based characterization of bullets was feasible in terms of size and shape but not composition.
