Image Handling & Central Review
Images leave the site de-identified within two working days, land in one encrypted central repository, and are read by two independent, blinded central readers who each lock within five working days. Comparison happens only after both locks. The clinical radiology read is never held for any research step.
A
Upload from the site (within 2 working days)
| What is uploaded | Notes |
|---|---|
| Complete TCUS still and cine dataset | All views, both probes |
| Calibration or tracking data | When volumetry was attempted |
| Full CT/MRI DICOM series | Whole-brain source images, not screenshots |
| Official radiology report | As signed |
| Identifiers on each upload | Study ID, window/side, encounter identifier, operator name Under review |
- De-identification: embedded identifiers removed using research PACS settings or anonymization tools; dates may be converted to study days. The site keeps the identifying link.
- Destination: the mandatory, encrypted central imaging-core repository with role-based access, transferred over approved secure channels under data-use agreements. Hospital PACS copies do not replace central submission. Repository platform Under review.
- The core data manager verifies readability, spatial calibration, completeness, and correct pairing, and queries the site for missing or mismatched files.
B
Central TCUS read (blinded to CT/MRI; locks within 5 working days)
- Reader: a central reviewer with documented TCUS competency, assigned by the imaging-core director, not involved in the patient's care. Andrei Alexandrov, MD — pending confirmation.
- Receives: Study ID, operated side, procedure type, postoperative interval, permitted prior TCUS studies. Never receives the paired CT/MRI images, report, or management decisions.
- Records: visible residual SDH boundaries and volume (manual contour on calibrated planes; mL, side, method, confidence); the selected midline landmark and distance; quality and coverage limitations; and a management impression that states whether the partial view is sufficient or formal imaging is required.
- Distinguishes SDH from postoperative air, CSF-density hygroma, and other fluid; uncertain boundaries are marked indeterminate.
C
Reference read (blinded to TCUS; locks within 5 working days)
- The routine clinical radiologist reports the CT/MRI exactly as usual, without waiting for anything.
- Site staff abstract the official midline shift (magnitude, direction, landmark, timestamp) as written.
- A separate central neuroradiology reader, blinded to all TCUS data, segments total residual SDH, records a standardized midline-shift measurement (septum pellucidum displacement perpendicular to the ideal midsagittal line at the foramen-of-Monro level), and records a CT/MRI-based research management impression. Elizabeth Westfall, MD — pending confirmation.
- The core measurement never silently replaces the official report value.
D
Post-lock matching and comparison
- A designated core analyst registers the documented TCUS field to the CT/MRI using the implant/window location, acquisition geometry, and anatomical landmarks; saves the transformation and overlay; clips the locked whole-brain segmentation to the TCUS region. Neither original read is changed.
- If the region cannot be localized reliably, the matched volume is non-evaluable; it is never estimated from an assumed field.
- The coordinating-center research team compares the TCUS impression with the CT/MRI research impression and classifies additional findings as inside, outside, or of uncertain relationship to the TCUS field.
Blinding safeguards
- Separate database permissions and image-access controls, not delays to clinical reads.
- Every reader attests to prior exposure; an exposed TCUS reader is replaced by a blinded backup or the pair is flagged as a blinding deviation.
- Corrections require an authorized, audited query; original value, reason, author, and time stay visible. Post-unblinding adjudication never overwrites locked primary values.
Adjudication (limited by design)
- Triggered only when (1) TCUS and CT/MRI interpretations are discordant, or (2) CT/MRI changes management despite a reassuring TCUS impression.
- Two independent investigators review the pair; a third resolves disagreement. No external adjudication panel.
Quality control
- A prospectively selected random 20% of evaluable pairs receive a second independent blinded TCUS and reference read for inter-reader reliability; second reads are retained separately.
- The core provides site feedback and requires retraining where acquisition standards are not met.
- Operator-level variability within each site is tracked from the operator name on each exam. Under review
Data system
REDCap or an equivalent 21 CFR Part 11-compliant system with role-based access, range checks, and source verification. Independent TCUS and CT/MRI research forms stay mutually concealed until both are locked. Only coded data enter the analysis dataset.