Abstract : Reverse transcription polymerase chain reaction (RT-PCR) is the gold standard for the diagnosis of COVID-19. Studies have proven that non-invasive techniques based on medical imaging can be used as an alternative to RT-PCR. The use of medical imag- ing technologies along with RT-PCR could improve the diagnosis and management of the disease. Even though several methods exist for diagnosing COVID-19 from X- ray images and CT scans, ultrasound image has not been explored much to diagnose the disease. In this study, we built a deep learning model using ultrasound images for a fast and efficient disease diagnosis. Pre-trained Convolutional Neural Networks (CNN), trained on the ImageNet database has been utilized for feature extraction. The nature-inspired Manta Ray Foraging Optimization (MRFO) algorithm is applied for dimensionality reduction and K-Nearest-Neighbour (KNN) for classification. Model training has been performed using a publicly available POCUS dataset consisting of 2944 ultrasound images sampled from more than 200 Lung Ultrasound (LUS) videos. Experimentations conducted in this study prove the efficiency of the model in the diagnosis of COVID-19. The model achieved an accuracy of 99.4337% using MobilenetV2 as the pre-trained network.
Alan : Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|