Related Experiment Video
Updated: Oct 14, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Fully automated segmentation in temporal bone CT with neural network: a preliminary assessment study.
Jiang Wang1, Yi Lv2, Junchen Wang2
1Department of Otorhinolaryngology-Head and Neck Surgery, Peking University Third Hospital, Peking University, NO. 49 North Garden Road, Haidian District, Beijing, 100191, China.
BMC Medical Imaging
|November 10, 2021
Summary
A new deep learning model automates temporal bone CT segmentation, crucial for image-guided otologic surgery. This AI approach shows promise for surgical planning and education.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Otolaryngology
Background:
- Manual segmentation of temporal bone CT scans is time-consuming.
- Accurate segmentation of temporal bone structures is essential for image-guided otologic surgery.
Purpose of the Study:
- To assess the feasibility and generalization ability of a deep learning model for automated segmentation of critical temporal bone CT structures.
Main Methods:
- A deep learning model was developed and tested on 39 temporal bone CT volumes (58 ears).
- Data included normal and abnormal cases with conditions like ossicular chain disruption and Mondini dysplasia.
- Segmentation accuracy was evaluated using Dice coefficient (DC) and average symmetric surface distance (ASSD).
Main Results:
- In normal cases, the model achieved high accuracy (e.g., DC of 0.910 for labyrinth).
- In abnormal cases, the model demonstrated moderate to good performance (e.g., DC of 0.698 for aberrant ossicles).
- The model showed good generalization across different temporal bone structures and pathologies.
Conclusions:
- The proposed deep learning model offers a feasible and generalizable solution for automated temporal bone CT segmentation.
- This technology holds significant potential for improving otologist education, disease diagnosis, and preoperative planning in image-guided otologic surgery.
