DentalSegmentator — NIfTI pipeline (user-uploaded CBCT)

Full end-to-end pipeline: drop your own .nii.gz CBCT in, WASM does gzip decompress + nnU-Net preprocess (transpose + trilinear resample to 0.432 × 0.312 × 0.432 mm + CTNormalization) + sliding window + inverse resample back to your scanner frame + NIfTI pack, and hands you back a labelled seg.nii.gz.  ·  ← single-patch parity/debug
How to use: ① Load WASM → ② Load model weights → ③ Upload a CBCT volume (NIfTI, gzipped or raw) → ④ Run sliding window → ⑤ Download segmentation.
✓ nnU-Net V2 strict conformance: end-to-end 99.9934% argmax agreement with nnUNetv2_predict (CPU, fold 0, use_mirroring=False). Pipeline order matches official exactly: transpose → crop-to-nonzero → CTNormalization (fingerprint μ/σ + clip to 0.5/99.5 percentiles) → cubic B-spline resample → sliding window. Cubic B-spline matches scipy bit-for-bit; postprocess resamples Ncls logits and argmaxes in the cropped frame, then embeds back into the full NIfTI volume.

Log

Axial slice viewer

Canonical input (CT-normalized)
Predicted segmentation (argmax)
Overlay