Local binary pattern and deep learning feature extraction fusion for COVID-19 detection on computed tomography...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• The COVID-19 pandemic necessitates rapid and accurate diagnostic methods.
• Traditional RT-PCR testing faces limitations due to kit scarcity and complexity.
• Medical imaging with machine learning offers a promising alternative for respiratory disease detection.

Purpose of the Study:

• To enhance COVID-19 detection accuracy using hybrid feature extraction techniques.
• To evaluate the performance of Support Vector Machines (SVM) and K-nearest neighbor (KNN) classifiers with combined features.
• To compare the proposed method against existing state-of-the-art models.

Main Methods:

• Handcrafted Local Binary Pattern (LBP) features were extracted.
• Features were automatically extracted using seven deep learning models.
• A hybrid approach concatenating LBP and deep learning features was proposed.
• SVM and KNN classifiers were trained using both individual and hybrid features.

Main Results:

• The VGG-19 + LBP model achieved a highest accuracy of 99.4%.
• Classifiers trained on the concatenated LBP and deep learning features significantly outperformed those trained on individual features.
• The hybrid feature approach demonstrated superior performance compared to state-of-the-art models.

Conclusions:

• The proposed hybrid feature extraction method significantly improves classifier performance for COVID-19 detection.
• Combining handcrafted and deep learning features offers a robust approach for medical image analysis in diagnostics.
• This technique holds potential for more efficient and accurate early detection of COVID-19.