Acute brain injury (ABI) is a leading cause of death and long-term disability in children. Diagnosis depends on fast neuroimaging, and head computed tomography (CT) is the bedside reference standard despite exposing the developing brain to ionizing radiation. A safe, portable, radiation-free bedside tool is needed to help decide which children need a head CT, to speed CT when injury is present, and to safely monitor injured children between scans. This study evaluates B-mode cranial point-of-care ultrasound (cPOCUS) performed through the temporal (and frontal) bone windows in children undergoing a clinically indicated head CT in the pediatric emergency department (ED) and pediatric intensive care unit (PICU). Children's thin skulls provide adequate acoustic windows in more than 95% of cases. cPOCUS is acquired by trained research scanners within 6 hours of head CT and interpreted offline by two blinded expert readers; the radiology CT report is the diagnostic gold standard. This is a prospective observational study; cPOCUS results are not returned to the clinical team in real time and do not change clinical care.
Prospective observational cohort study of children undergoing clinically indicated head CT for suspected acute brain injury (ABI) in the pediatric ED and PICU at the Children's Hospital of Richmond at Virginia Commonwealth University (CHoR-VCU). Within 6 hours of the head CT, trained research scanners (acquisition-only role) acquire standardized B-mode cranial point-of-care ultrasound (cPOCUS) using 1-3 MHz phased-array probes through bilateral temporal and frontal windows, exporting de-identified 3-5 second cine loops at and around the midbrain plane to an encrypted offline reading workstation. Two pre-specified expert readers interpret the studies independently and offline in Horos, each blinded to the clinical course, the head-CT report, and one another; discordant cases are resolved by joint re-read to a single adjudicated consensus interpretation, which serves as the cPOCUS index test. The attending radiology head-CT report is the diagnostic gold (reference) standard. Because reads are not returned to the clinical team in real time, cPOCUS is not an assigned intervention, and the endpoints are diagnostic test characteristics, there is no change to clinical care.
Study Type
OBSERVATIONAL
Enrollment
220
Virginia Commonwealth University
Richmond, Virginia, United States
Diagnostic accuracy of cranial point-of-care ultrasound (cPOCUS) versus head CT
Diagnostic accuracy of the adjudicated consensus cPOCUS read (index test) for a binary abnormal versus normal head CT (reference standard = attending radiology CT report), reported as sensitivity, specificity, positive predictive value, and negative predictive value, each with 95% confidence intervals, at the first paired CT-cPOCUS assessment. Abnormal scan is defined as presence of any new midline shift, hydrocephalus or parenchymal echogenicity.
Time frame: First paired CT-cPOCUS assessment (cPOCUS within 6 hours of head CT; Day 1)
Inter-reader reliability of cPOCUS interpretation
Agreement between the two blinded expert readers on the binary cPOCUS classification of midline shift, hydrocephalus or parenchymal echogenicity, quantified by Cohen's kappa (prespecified threshold kappa \>= 0.8); observed agreement and Gwet's AC1 reported given anticipated low prevalence.
Time frame: First paired assessment; monitored quarterly through study completion (up to 24 months)
Per-reader diagnostic accuracy and paired reader comparison
Each reader's independent binary cPOCUS call scored against head CT for per-reader sensitivity and specificity for midline shift, hydrocephalus or parenchymal echogenicity ; the two readers compared on the same patients by the paired McNemar test.
Time frame: First paired CT-cPOCUS assessment (Day 1)
Etiology- and subgroup-specific diagnostic accuracy (descriptive)
Descriptive, explicitly underpowered estimates of cPOCUS accuracy by prespecified subgroups: age band (2-5, 6-12, 13-17 years), acoustic-window adequacy, mechanism/clinical presentation, and underlying diagnosis; sex-stratified accuracy prespecified per NIH sex-as-a-biological-variable policy.
Time frame: First paired CT-cPOCUS assessment (Day 1)
Age and skull-thickness inflection point (upper age limit of utility)
Identification of the age/skull-thickness threshold above which cranial ultrasound loses screening utility, from age-stratified accuracy and window-adequacy analysis.
Time frame: Through study completion (up to 24 months)
Pediatric cPOCUS atlas feature and artifact agreement
Inter- and intra-observer agreement (Cohen's kappa, age-stratified) on predefined B-mode features and artifacts annotated with a structured ontology, and CT spatial concordance by Horos reformatting/transparency fusion. Atlas and ontology released open access (Zenodo DOI).
Time frame: Through study completion (up to 24 months)
Longitudinal midline-shift agreement (cPOCUS vs CT)
Agreement between interval change in midline shift (mm) on serial cPOCUS and serial head CT in the CT-positive PICU subset, assessed by Bland-Altman analysis with prespecified limits of agreement.
Time frame: Serial paired scans over ICU stay (up to 24 months)
Longitudinal hydrocephalus and new parenchymal echogenicity agreement
Paired concordance between serial cPOCUS and serial head CT for hydrocephalus (present/absent) and new parenchymal echogenicity/new lesion (present/absent), assessed by McNemar's test.
Time frame: Serial paired scans over ICU stay (up to 24 months)
Evolving versus stable pathology (exploratory longitudinal model)
Exploratory modeling of evolving versus stable pathology on the subsequent CT using logistic generalized estimating equations (GEE) with robust standard errors (patient as clustering unit); decision-curve analysis for net clinical benefit.
Time frame: Serial paired scans over ICU stay (up to 24 months)
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.