NeurobookMind · Skull Base Fellowship V8STRUCTURED FELLOWSHIP · ANATOMY FIRST
OFFLINE ENGINE · LOCAL DATA
Advanced extended endoscopic endonasal approach

Image-derived skull-base anatomy, not a fabricated 3D skull.

Inspect authentic CT voxels in three planes, choose your sphenoid/sellar region of interest, reconstruct a continuous triangulated bone surface, and confirm all boundaries against the original slices. Add MRI separately for soft-tissue interpretation.

POST-BOARD · THEORY & IMAGING LAB
Initializing image controls…
Post-board · structured surgical fellowship

Advanced Skull Base Fellowship

Learn in sequence: anatomy → vascularization → corridors → indications → operative technique → reconstruction → complications → imaging integration → clinical decision-making. Cases and imaging labs support the curriculum; they do not replace foundational teaching.

Current learning stage: Module 01 · Nasal cavity anatomy
Start with orientation and anatomical boundaries. Do not jump to CT/3D before the anatomy lesson.
ANATOMY FIRST

Module lessons

Learning sequence: complete the anatomy lesson and checkpoint before progressing to mucosa, vascular anatomy and surgical corridors. Imaging tools remain available as a separate laboratory when the relevant anatomy has been taught.
Step 01 · Source data

Imaging Laboratory · CT / MRI

This is a supporting imaging laboratory, not the starting point of the fellowship. The app processes files locally in Safari. Large uncompressed .nii files use low-memory streaming; compressed .nii.gz images may need extraction first. Import a real CT volume in NIfTI (.nii / .nii.gz) or attached-data NRRD (.nrrd) format. CT and MRI remain independent until professionally registered.

No CT or MRI imported. The reconstruction panel will stay empty until a volume is loaded; synthetic QA is explicitly labeled.
Large-CT streaming: uncompressed NIfTI (.nii) can be previewed without loading its entire payload. Native-resolution voxels are read only for the chosen 3D ROI. Use patient-free educational datasets.
AXIAL*—
Move slice / tap to relocate 3D ROI
CORONAL*—
Crosshair determines ROI center
SAGITTAL*—
Inspect sellar, clival boundaries

*Axis names are voxel-plane labels; NIfTI orientation transforms are reported but are not resampled to radiological orientation. Verify anatomical left/right and axis directions in a validated viewer before interpretation.

—Display voxel dimensions
—Voxel spacing (mm)
—Data source
Step 02 · CT surface reconstruction

Continuous 3D bone mesh · Marching tetrahedra

Real triangles are extracted from threshold-crossings in the loaded CT—not a point cloud. Rebuild at different thresholds and quality. This is a threshold-based surface, not verified anatomical segmentation; inspect all suspect defects in the original CT.

3D VIEW · WAITING FOR CTONE FINGER: ROTATE · PINCH: ZOOM · BUTTONS: ROTATE / RESET
ROI & extraction controls
350
90 mm
Set the ROI center by tapping the relevant anatomy on the CT slices; the 3D reconstruction uses that point.
Reconstruction status

No scan selected. Load CT, set your ROI, then build.

ROI center: —

Requires real thin-slice bone-kernel CT. A coarse or thick-slice study cannot recover thin sphenoid septa or optic canal dehiscence. Thresholding alone may still create false holes; segmentation validation is mandatory.
Step 03 · Anatomical mastery

EEEA imaging objectives

These checkpoints must be identified on real imaging. No automatic identification is claimed.

Expert teaching notes

Step 04 · Assessment

Imaging quality and approach planning checkpoint

0 / 5

Evidence and limitations

What this workstation can and cannot do

Working now: local NIfTI / attached-data NRRD loading, genuine orthogonal voxel views, isosurface triangle generation from CT intensities, touch orbit, zoom, ROI controls, image-quality flags and structured expert checkpoints.

Not implemented: validated sphenoid, carotid canal, optic nerve or sellar segmentation; same-patient CT–MRI registration; surgical navigation; DICOM series decoding; automated surgical measurements. No high-resolution clinical CT is embedded in this package.

Privacy: do not upload identifiable patient data to Cloudflare or ChatGPT. This version processes locally selected files in your browser; use properly de-identified, permissioned teaching imaging only.

Imaging targets for serious EEEA fellowship

Thin-slice CT bone kernel, generally around 0.5–0.625 mm or better when available, plus dedicated contrast-enhanced 3D T1 MRI and appropriate vascular imaging for selected approaches. Compare MRI and CT only after accurate registration and visual verification.

Reference: 3D endoscopic navigation & skull-base CT reconstruction. Study uses 0.6 mm bone-kernel CT and separately segmented osseous, ICA and optic structures.

Dataset note: CT-SCOPE includes 0.47–1 mm sinus CT data, but the Zenodo files are currently restricted and cannot be shipped here. NasalSeg provides openly downloadable paranasal CT, but its reported 1.5 mm slice spacing is inadequate to validate fine dehiscence. NasalSeg · CT-SCOPE.

Do not mistake a visually smooth surface for surgical accuracy. Accurate 3D reconstruction requires validated imaging, correct geometry/orientation, quality-controlled segmentation, and expert anatomical review.
NeurobookMind Skull Base Fellowship V8 · educational fellowship workspace · no server processing of imported scans · no clinical clearance or accreditation.