Fusion extracted features from deep learning for identification of multiple positioning errors in dental panoramic...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Dentistry
• Machine Learning Applications

Background:

• Dental panoramic imaging is crucial for diagnosis and treatment planning.
• Patient positioning errors in panoramic imaging can degrade image quality and lead to misdiagnosis.
• Accurate patient positioning is technically challenging due to equipment complexity and anatomical variations.

Purpose of the Study:

• To develop and validate a deep learning model for accurate and efficient identification of multiple positioning errors in dental panoramic imaging.
• To enhance the diagnostic accuracy and efficiency of dental panoramic imaging through automated error detection.

Main Methods:

• A retrospective study utilizing 552 dental panoramic images.
• Development of six Convolutional Neural Network (CNN) models for feature extraction, followed by feature fusion using transfer learning.
• Implementation of a Support Vector Machine (SVM) classifier with six binary SVMs to identify six defined positioning errors (slumped position, chin tipped low, open lip, head turned, head tilted, tongue against palate).

Main Results:

• The fusion of image features with six binary SVM classifiers demonstrated high performance in identifying multiple positioning errors.
• The developed classifier achieved an overall accuracy of 0.832 for detecting various positioning errors.
• The system showed high precision and recall rates in experimental evaluations.

Conclusions:

• The study successfully demonstrates the effectiveness of SVM classifiers in identifying multiple positioning errors in dental panoramic imaging.
• The integration of feature fusion and SVM classifiers enhances diagnostic precision and suggests improvements in dental imaging efficiency.
• Future research should focus on larger datasets and real-time clinical applications for broader implementation.