Related Experiment Video
Updated: Sep 11, 2025
07:06
Endaural Endoscopic Atticoantrotomy Retrograde Mastoidectomy using a Constant Suction Bone-drilling Technique
Published on: May 23, 2021
3.9K
Automated Prediction of Bone Volume Removed in Mastoidectomy
Nimesh V Nagururu1,2, Hisashi Ishida2, Andy S Ding1
1Department of Otolaryngology-Head and Neck Surgery, Johns Hopkins University, Baltimore, Maryland, USA.
Summary
This study introduces a deep learning method to predict bone removal during mastoidectomy, offering patient-specific surgical guidance. The AI model shows promise in approximating surgical endpoints, aiding training and computer-assisted surgery.
Area of Science:
- Neurosurgery
- Medical Imaging
- Artificial Intelligence
Background:
- Mastoidectomy bone volume is critical for surgical success and safety.
- Predicting bone removal aids surgical training and advanced interventions.
Purpose of the Study:
- To develop a deep learning pipeline for automated prediction of bone volume removed during mastoidectomy.
- To evaluate the accuracy of AI-driven predictions against actual surgical outcomes.
Main Methods:
- A deep learning pipeline was created using VR mastoidectomy simulation data.
- The dataset comprised 15 temporal bone CT scans.
- Model performance was assessed using Dice score (DSC) and Hausdorff distance (HD) via cross-validation.
Main Results:
- The method achieved a median DSC of 0.775 and a median HD of 0.492 mm.
- Predictions successfully identified key surgical endpoints in 80% of cases.
- Qualitative analysis revealed generally realistic, though occasionally imprecise, bone removal predictions.
Conclusions:
- Deep learning offers a viable approach for predicting bone volume in mastoidectomy.
- Current models provide a reasonable approximation of surgical endpoints.
- Further improvements require larger datasets and advanced model architectures.
