Artificial intelligence in detection and segmentation of internal auditory canal and its nerves using deep learning...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Deep Learning

Background:

• Artificial intelligence (AI) is increasingly vital in medical imaging for interpreting complex structures.
• Deep learning advancements enable powerful feature extraction without human intervention.
• Accurate detection and segmentation of subtle tissues like the internal auditory canal (IAC) remain challenging.

Purpose of the Study:

• To present an automated method for detecting and segmenting the IAC and its associated nerves (facial, cochlear, vestibular).
• To leverage deep learning techniques to improve the accuracy of IAC and nerve analysis.
• To develop an AI-driven tool to assist radiologists in clinical decision-making.

Main Methods:

• A Mask R-CNN approach, utilizing a RESNET50 backbone, was employed for IAC localization.
• A U-net model was integrated to segment the IAC and its nerves based on learned features.
• The combined Mask R-CNN and U-net architecture was trained and validated on clinical datasets.

Main Results:

• The Mask R-CNN approach achieved a mean Intersection of Union (IoU) of 0.79 for IAC localization with RESNET50.
• The U-net segmentation method yielded a Dice similarity coefficient of 96% for IAC and its nerves.
• The proposed AI method demonstrated superior performance in both localization and segmentation tasks compared to traditional methods.

Conclusions:

• The developed AI method accurately localizes and segments the internal auditory canal and its nerves.
• This AI-driven approach supports radiologists by providing precise anatomical information.
• The findings highlight the potential of AI in enhancing diagnostic capabilities in neuroimaging.