JawFracNet


Logo for JawFracNet

About

Creators:
Image Version:
fb8898b5-94b2-43a0-a638-b53fe41f3d39
Last updated:
Dec. 15, 2022, 8:26 a.m.
Inputs:
  • CT Image  (Any CT image)
Outputs:
  • Mandible segmentation  (Segmentation of the mandible with 0 = background, 1 = mandible)
  • Fractures  (Segmentation of detected fractures with 0 = background and 1=fracture)

Model Facts

Summary

JawFracNet processes 3D volume patches sampled from the input CBCT scan to predict a mandible segmentation and mandibular fracture segmentations.

The image below provides an example.

Mechanism

Target population

The algorithm targets patients with trauma to the head for which a mandibular fracture is suspected.

Algorithm description

JawFracNet is implemented with a three-stage model, see the figure below.

3D volume patches are sampled from the input scan and processed by the model.

Stage 1 predicts a mandible segmentation in each patch and aggregates the patch predictions for the final mandible segmentation.

Stage 2 gets contextual features from stage 1 and predicts fracture segmentations in the patch.

Stage 3 predicts whether the patch contains a fracture with binary classification.

The results from stage 2 and stage 3 are combined for the final fracture segmentations.

Input

The input is a head CBCT scan. JawFracNet is trained and evaluated only on head CBCT scans, so it is not effective for scans that include other regions as well.

Outputs

The outputs are the final mandible and fracture segmentations.

Validation and Performance

Uses and Directions

This algorithm was developed for research purposes only.

Warnings

Common Error Messages

Information on this algorithm has been provided by the Algorithm Editors, following the Model Facts labels guidelines from Sendak, M.P., Gao, M., Brajer, N. et al. Presenting machine learning model information to clinical end users with model facts labels. npj Digit. Med. 3, 41 (2020). 10.1038/s41746-020-0253-3