Texture-based speciation of otitis media-related bacterial biofilms from optical coherence tomography images using...

Area of Science:

• Biomedical Engineering
• Infectious Diseases
• Medical Imaging

Background:

• Otitis media (OM) is a common childhood ear infection, often caused by bacterial biofilms.
• Recurrent or chronic OM can involve antibiotic-resistant bacterial species.
• Optical coherence tomography (OCT) visualizes middle ear biofilms.

Purpose of the Study:

• To compare microstructural image texture features of bacterial biofilms using OCT.
• To develop a machine learning framework for classifying bacterial species within OM biofilms.
• To assess the potential for real-time in vivo characterization of ear infections.

Main Methods:

• Applied supervised machine learning (SVM, random forest, XGBoost) to OCT images.
• Analyzed texture features from in vitro bacterial cultures and in vivo human subject images.
• Optimized Support Vector Machine with Radial Basis Function (SVM-RBF) and XGBoost classifiers.

Main Results:

• Classifiers achieved over 95% Area Under the Curve (AUC) for detecting biofilm classes.
• Demonstrated high accuracy in differentiating multiple bacterial species.
• Validated the approach on both laboratory cultures and clinical patient data.

Conclusions:

• Texture analysis of OCT images combined with machine learning can differentiate OM-causing bacterial biofilms.
• This approach shows promise for accurate, real-time in vivo diagnosis of ear infections.
• Offers potential for improved treatment strategies for otitis media.