Endpoints & Study Goals

The primary question is whether the partial view through the implant window agrees with whole-brain imaging in the same region, and whether movement of visible midline structures tracks official midline shift. Diagnostic accuracy and management impact are secondary and tertiary. This is not a non-inferiority or replacement trial.

Primary endpoint — Specific Aim 1: Partial brain information in relation to whole-brain imaging

Principal quantitative endpoint: residual SDH volume (mL) within the predefined, documented TCUS viewing region, compared with CT/MRI segmentation restricted to that same region. Change from the first evaluable paired exam is analyzed where a reproducible common region exists.
Complementary component: distance from a fixed point on the inner implant surface to a midline landmark (septum pellucidum; center of the third ventricle as a labeled fallback), its serial change, and its association with the official CT/MRI midline-shift measurement and a standardized core measurement.
Blood outside the TCUS region is quantified separately on CT/MRI and is not added to the matched-region comparator.
Ventricular size and ventricular-horn measurements are not required Aim 1 endpoints.
Analysis: bias, absolute error, intraclass correlation, and Bland–Altman limits of agreement, accounting for repeated encounters within patients. No universal ±5% criterion; clinically acceptable error thresholds are set in the statistical analysis plan before enrollment. Under review

Secondary endpoints

Safety and feasibility: implant- and ultrasound-related complications (wound issues, scalp irritation, bruising, any delay in care attributable to TCUS) over the follow-up period.
Image quality: distribution of exam grades (Good / Limited / Non-diagnostic), factors associated with suboptimal images, and the learning curve over time.
Diagnostic accuracy for prespecified findings on whole-brain reference imaging: (a) new or enlarging SDH within the documented TCUS field, and (b) worsening mass effect or official midline shift > 5 mm. Reported as sensitivity, specificity, predictive values, and accuracy with patient-clustered 95% confidence intervals. Indeterminate and non-diagnostic exams get their own denominator plus a conservative analysis counting them as test failures. A limited field is never scored as a negative whole-brain study.
Site-to-central correlation: blinded and unblinded correlation between each site's bedside measurements and the central imaging core's locked measurements. Under review — requested by Dr. Nimjee; not yet written into the protocol.

Tertiary endpoint — Clinical management impact (hybrid endpoint)

Proportion of paired encounters in which the imaging-based management impression is unchanged after whole-brain CT/MRI review, relative to the impression recorded from TCUS alone.
Counts only when the TCUS reader first judged the view sufficient and did not recommend additional imaging. Non-diagnostic or insufficient exams cannot qualify.
Additional whole-brain findings are classified as inside, outside, or of uncertain relationship to the TCUS field. Diagnostic discordance, changed research impressions, and actual clinical actions are reported separately.
Any "potentially avoidable scan" classification is hypothetical. No CT or MRI is withheld during this study.

What the study produces

A Level 2 diagnostic-evidence dataset defining both the useful information and the coverage limits of partial brain imaging as an adjunct to postoperative SDH surveillance.
A safety and feasibility review at ~50% enrollment (50–60 patients) covering acquisition quality, volumetry feasibility, landmark visibility, registration failure, protocol adherence, and recruitment.
Peer-reviewed publication reported to STARD 2015 standards. De-identified data may be shared on reasonable request after publication with IRB approval.

What the study does not claim

It will not establish that TCUS can replace whole-brain imaging for any patient or encounter. A reassuring partial view cannot be assumed to exclude pathology elsewhere.