# Benchmark

With every release, we run a benchmark of the most important algorithms and segmentations. The benchmark is calculated using a large dicom (569.1 MB) and two IOSes (18.8 MB & 20.5 MB). You can find the results for every release inside the changelog.

# Explanation of benchmark types

For an in-depth explanation of what each algorithm does, take a look at the key functionality.

• All IOS algorithms are ran on two jaws.
• Full virtual patient: All computations we do on a creator job. It includes:
  • Teeth, Nerves, Mandible, Skull, Maxillary, Sinusses and airway segmentations in CBCT
  • Teeth segmentations in IOS
  • Registration between CBCT and IOS
  • Registration between Facial scan and CBCT
  • Fused teeth
  • Closed crowns
  • Panoramic arch generation
  • Teeth bounding boxes
  • Cleaned IOS (with base)
  • IOS landmarks
• Full IOS: All computations we do on a creator job, when only an IOS is provided. It includes:
  • Teeth segmentations in IOS
  • Closed crowns
  • Cleaned IOS (with base)
  • IOS landmarks