What a voxel actually is
A voxel is the 3D equivalent of a pixel. CBCT produces an isotropic volume — voxels are cubic, so the same resolution applies in X, Y and Z. A 0.2 mm voxel is 200 microns on each side.
Reconstructed voxel size is set by the operator when choosing a scan protocol, and is constrained by the physical detector pitch. Finer voxels than the detector supports are interpolated, not truly resolved.
Trade-offs of smaller voxels
Higher detail resolves fine structures like accessory canals, root fractures and fenestrations of the cortical plate. This is crucial in endodontic re-treatment and medico-legal cases.
Cost: scan time is longer, patient motion more likely, dose rises (often linearly with 1/voxel³ for comparable SNR), and noise becomes more visible. If the indication does not need sub-millimetre detail, smaller voxels add dose without clinical benefit.
Typical voxel size per indication
Endodontics and root fractures: 0.075 to 0.15 mm. Often with a small FOV (4 × 4 cm or 5 × 5 cm) to limit dose.
Implant planning: 0.2 to 0.3 mm. A FOV covering one arch or both jaws is usual.
Orthodontics and orthognathic: 0.25 to 0.3 mm. Full-skull or hemi-cranial FOV.
Airway, TMJ, impacted teeth survey: 0.3 to 0.4 mm. Dose and SNR matter more than sub-millimetre detail.
What viewers do with voxel data
Any modern CBCT viewer reads the DICOM header and respects the voxel spacing when rendering axial, coronal and sagittal slices, when performing measurements, and when building MPR and 3D reconstructions. If the viewer ignores voxel spacing, measurements will be wrong.
High-quality viewers also offer resampling — temporarily rendering thicker slabs for review speed, or thinner for fine detail. This does not change the underlying voxel size.