An in-house fabricated patient-specific pediatric head CT phantom for trauma-oriented emergency imaging: two-scanner...

To evaluate an in-house newly fabricated patient-specific pediatric head CT phantom developed for trauma-oriented emergency imaging, with emphasis on multiplanar visual realism, cross-platform interpretability, and expert-reader assessment of diagnostic acceptability.

A heterogeneous phantom was generated from an anonymized retrospective head CT dataset of a 5-year-old child and designed to preserve emergency-relevant osseous, aerated, orbital, and intracranial anatomy. A complex orbito-nasal midface trauma was incorporated as a trauma-mimicking feature. CT was performed on two clinical multidetector systems from different vendors: a 64-row GE Discovery CT750 HD and an 80-row Canon Aquilion RXL, each using routine pediatric head CT protocols. Ten radiologists independently reviewed axial, coronal, and sagittal reformations in bone and soft-tissue windows using a structured 20-item, 4-domain scoring framework.

Representative multiplanar images showed preserved calvarial contour, skull-base and orbital anatomy, aerated sinonasal spaces, intracranial contour, falcine/midline anatomy, ventricular depiction, posterior fossa appearance, and visualization of the complex orbito-nasal midface trauma. Expert ratings were favorable on both scanners, with higher scores on Canon across all domains. Domain means on Canon versus GE were 4.38 versus 4.02 for osseous realism, 4.32 versus 4.00 for soft-tissue realism, 4.36 versus 4.18 for multiplanar interpretability, and 4.57 versus 4.10 for clinical utility. Overall composite scores were 4.38 on Canon and 4.07 on GE, paired domain-level differences showed higher mean scores for the Canon dataset in all domains, with the largest difference in clinical utility (+ 0.47).

The phantom provided clinically recognizable, trauma-relevant pediatric emergency head CT appearances across multiplanar bone and soft-tissue review and was rated as realistic and diagnostically acceptable by expert readers, supporting its use in protocol evaluation, training, and observer-based emergency radiology research.