Related Experiment Video
Updated: Dec 27, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
Multi-Scale deep learning framework for cochlea localization, segmentation and analysis on clinical
Floris Heutink1, Valentin Koch2, Berit Verbist3
1Department of Otorhinolaryngology and Donders Institute for Brain, Cognition and Behavior, Radboudumc, Geert Grooteplein Zuid 10, 6525 GA, Nijmegen, the Netherlands.
Computer Methods and Programs in Biomedicine
|February 29, 2020
Summary
This study presents an automated deep learning method for segmenting and measuring the human cochlea from CT scans. The accurate measurements can aid in personalized cochlear implant surgery planning.
Area of Science:
- Medical Imaging
- Deep Learning
- Anatomy
Background:
- Patient-specific pre-operative cochlea measurements can improve cochlear surgery outcomes.
- Reducing intracochlear trauma and preserving residual hearing are key surgical goals.
Purpose of the Study:
- To develop an automated method for segmenting and measuring the human cochlea in ultra-high-resolution (UHR) CT images.
- To investigate cochlea size variations for personalized cochlear implant planning.
Main Methods:
- A deep learning system was developed and validated on 123 temporal bone CT scans.
- The system uses a cascade of detection and pixel-wise classification (U-Net-like architecture) for segmentation.
- Convolutional neural networks and thinning algorithms provide automatic measurements of cochlear volume, basal diameter, and cochlear duct length (CDL).
Main Results:
- The automated segmentation achieved high accuracy (Dice: 0.90, BF score: 0.95).
- Measurement errors were 8.4% for volume, 5.5% for CDL, and 7.8% for basal diameter.
- Significant variations in cochlea size were observed across the patient cohort.
Conclusions:
- The developed algorithm accurately segments and measures cochlear anatomy on UHR-CT images.
- The findings support the use of this method as a pre-operative tool for personalized cochlear implant strategies.
