Presentation Category
Research
Introduction/Context/Diagnosis
To evaluate the accuracy of a fully automated landmark identification (ALI) system as a tool for automatic landmark location compared with human judges. Materials and
Methods/Treatment Plan
Methods: A total of 76 cone-beam computed tomography (CBCT) images were collected. After the calibration procedure, two human judges identified 28 landmarks (17 hard tissue, 4 dental, 7 soft tissue) in the x, y, and z coordinate planes on CBCTs using Invivo software. The ground truth was created by averaging landmark coordinates identified by two human judges for each landmark. To evaluate the accuracy of ALI, the mean absolute error (mm) at the x, y, and z coordinates and mean error distance (mm) between the human landmark identification and the ALI were determined. A successful detection rate was calculated.
Results/Outcome
The ALI's mean absolute error for all coordinates was 1.6 mm on average. Across all three coordinate planes, 99% of the landmarks had a mean absolute error of less than 3 mm. The mean error distance for all 28 landmarks was 3.27 ± 1.9 mm. When applied to 28 landmarks on 76 CBCTs, the ALI system showed a 73% success rate in detecting landmarks within a 4-mm error distance range.
Significance/Conclusions
Overall, ALI showed clinically acceptable mean error distances except for a few landmarks. This study demonstrates the promise of ALI in aiding orthodontists with landmark identifications on CBCTs.
Format
Event
Accuracy of 3-D automated landmark identification on cone-beam computed tomography
To evaluate the accuracy of a fully automated landmark identification (ALI) system as a tool for automatic landmark location compared with human judges. Materials and
Comments/Acknowledgements
Richa Roy, Young Eun Jung, Hanisha Pasupulate, Heeyeon Suh, Joorok Park, Heesoo Oh