Introduction: We sought to validate a handheld, near-infrared spectroscopy (NIRS) device for detecting intracranial hematomas in children with head injury.
Methods: Eligible patients were those <18 years old who were admitted to the emergency department at three academic children's hospitals with head trauma and who received a clinically indicated head computed tomography (HCT). Measurements were obtained by a blinded operator in bilateral frontal, temporal, parietal, and occipital regions. Qualifying hematomas were a priori determined to be within the brain scanner's detection limits of >3.5 milliliters in volume and <2.5 centimeters from the surface of the brain. The device's measurements were positive if the difference in optical density between hemispheres was >0.2 on three successive scans. We calculated diagnostic performance measures with corresponding exact two-sided 95% Clopper-Pearson confidence intervals (CI). Hypothesis test evaluated whether predictive performance exceeded chance agreement (predictive Youden's index > 0).
Results: A total of 464 patients were enrolled and 344 met inclusion for primary data analysis: 10.5% (36/344) had evidence of a hematoma on HCT, and 4.7% (16/344) had qualifying hematomas. The handheld brain scanner demonstrated a sensitivity of 58.3% (21/36) and specificity of 67.9% (209/308) for hematomas of any size. For qualifying hematomas the scanner was designed to detect, sensitivity was 81% (13/16) and specificity was 67.4% (221/328). Predictive performance exceeded chance agreement with a predictive Youden's index of 0.11 (95% CI, 0.10 - 0.15; P < 0.001) for all hematomas, and 0.09 (95% CI, 0.08 - 0.12; P < 0.001) for qualifying hematomas.
Conclusion: The handheld brain scanner can non-invasively detect a subset of intracranial hematomas in children and may serve an adjunctive role to head-injury neuroimaging decision rules that predict the risk of clinically significant intracranial pathology after head trauma.