Artificial intelligence for otosclerosis detection on temporal bone CT: a diagnostic study.

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence

Background:

• Otosclerosis is a bone disorder affecting the otic capsule, leading to hearing loss.
• High-resolution computed tomography (HRCT) is crucial for diagnosis, but its accuracy depends on interpretation.
• Artificial intelligence (AI) offers potential to enhance the detection of subtle otosclerosis features.

Purpose of the Study:

• To assess the diagnostic accuracy of a convolutional neural network (CNN) for identifying otosclerosis.
• To evaluate the CNN's performance in detecting otosclerosis on temporal bone CT scans.

Main Methods:

• A retrospective study utilized CT scans from 53 otosclerosis patients and 36 controls.
• A CNN model was developed and trained on an augmented dataset of 74 otosclerosis and 74 control images.
• Performance metrics included accuracy, sensitivity, specificity, precision, F1-score, and AUC.

Main Results:

• The CNN achieved 98% training accuracy and 87.5% maximum validation accuracy.
• Validation metrics showed 80.0% sensitivity, 84.6% specificity, 85.7% precision, and an 0.847 AUC.
• Learning curves indicated stable model convergence without overfitting.

Conclusions:

• AI-powered CT analysis demonstrates significant potential for diagnosing otosclerosis, particularly in subtle or normal-appearing cases.
• Integrating AI into otologic imaging can improve diagnostic reliability and patient management.
• AI may enhance surgical planning for otosclerosis treatment.