Intraoral scan quality for aligner planning

Almost every case that has to be sent back is sent back for the scan. The prescription can be clarified in a message and an objective can be renegotiated, but a mesh with a hole where the contact should be cannot be planned around, because the planner would be inventing the anatomy that is missing. Five minutes of checking before submission prevents nearly all of it.

What actually makes a scan unusable

• Missing distal surfaces on the terminal molars, so the arch cannot be closed and the last tooth cannot be segmented.
• Holes at the interproximal contacts, which is where segmentation has to cut and where IPR will be measured.
• Soft tissue captured over the teeth: a cheek, a tongue or a finger merged into the arch as if it were anatomy.
• A bite scan that will not seat on the arch scans, usually because an arch was rescanned afterwards.
• Heavy decimation or smoothing on export, which rounds off cusps, marginal ridges and contact points.
• Warped stitching, where a long scan path has accumulated error and the arch is subtly the wrong shape.

Some of these are obvious and some are not. A hole is visible. A stitching error is not: the mesh looks complete, plans normally, and produces aligners that do not fit across the full arch. It is worth knowing that the most damaging scan defect is the one that does not look like a defect.

Capture habits that avoid the problem

Dry the teeth properly before starting and keep them dry, particularly at the contacts and around the gingival margin, since saliva and blood are the usual cause of a hole. Retract the cheek and the tongue actively rather than hoping. Follow a consistent scan path — occlusal, then lingual, then buccal, or whatever your scanner's manufacturer recommends — instead of returning repeatedly to patch small areas, because repeated re-entry is what accumulates registration error. Start at a terminal molar and finish there, so the hardest area is captured while your access is best.

Scan the bite with the patient closed in their habitual position and check on screen that the teeth are actually in contact, not held apart by the scanner tip or by the patient posturing. Capture both sides. And take the bite last, after both arches are final; if you then rescan an arch, retake the bite.

Scan the arch before anything is placed that will change it. A scan taken with a retraction cord, a wedge, cotton rolls in the buccal sulcus or an isolation device in place records those objects as anatomy, and the arch has to be scanned again.

Checking before you upload

1. Look at the model from the occlusal view of each arch and confirm every tooth is complete, including the last molar distally.
2. Rotate to the lingual and buccal and look along the gingival margin for gaps and torn edges.
3. Zoom into the contacts in the crowded regions, which is where holes hide.
4. Seat the bite and check that the contacts on screen match what you saw in the mouth.
5. Confirm you are exporting at full resolution, not a reduced or lightweight setting.
6. Open the exported file once before sending, so you know what the planner will receive.

That last step catches more problems than the rest combined, because what the scanner shows on its own screen and what it writes into an exported mesh are not always the same thing.

Why a small defect matters more than it looks

It is tempting to assume a planner can patch a hole, and for a small void in the middle of a smooth buccal surface that is broadly true. The places where it is not true are precisely the places scans tend to fail. A gap at an interproximal contact is where segmentation cuts, so the tooth boundary has to be guessed and the IPR measurement rests on invented anatomy. A missing distal surface on a terminal molar leaves the tooth without the reference the aligner will grip. A smoothed cusp tip removes the occlusal contact the plan is supposed to be checking. In each case the resulting model looks plausible and is wrong in the specific region the plan depends on.

Difficult cases

Some mouths make good scans hard: severe crowding where contacts are inaccessible, deep undercuts, mobile teeth that shift during capture, large restorations with reflective surfaces, and patients who cannot keep still or open long enough. For these, capture in shorter segments with clean overlap rather than one long pass, take extra care over drying, and consider scanning a difficult region twice and keeping the better result.

Where the anatomy simply cannot be captured — an unerupted tooth, a subgingival margin, a region behind an appliance — say so in the case instructions. A planner told about a limitation can work around it explicitly. A planner who finds it unannounced has to stop and ask.

When a case is sent back

A rescan request should tell you what is wrong, where, and what to do about it: which arch, which region, whether the whole scan or one segment needs repeating. "Poor quality scan" is not an actionable message, and a planning partner that sends it is costing you a patient appointment you may not need. Equally, if a defect is genuinely minor and can be handled by interpolating a small region, the plan should say that it did so rather than silently filling it in.

The cost of a rescan is not the scan; it is bringing the patient back. That asymmetry is the whole reason for checking at the chair while they are still in it.