A Hybrid Automated Intelligent COVID-19 Classification System Based on Neutrosophic Logic and Machine Learning Techniques Using Chest X-Ray Images

Yasser, I.; Abd El-Khalek, A. A.; Twakol, A.; Abo-Elsoud, M. E.; Salama, A. A.; Khalifa, F.

To facilitate timely treatment and management of COVID-ap patients, efficient and quick identification of COVID-19 patients is of immense importance during the COVID-19 crisis. Technological developments in machine learning (ML) methods, edge computing, computer-aided medical diagnostic been utilized for COVID-19 Classification. This is mainly because of their ability to deal with Big data and their inherent robustness and ability to provide distinct output characteristics attributed to the underlying application. The contrary transcription-polymerase chain reaction is currently the clinical typical for COVID-19 diagnosis. Besides being expensive, it has low sensitivity and requires expert medical personnel. Compared with RT-PCR, chest X-rays are easily accessible with highly available annotated datasets and can be utilized as an ascendant alternative in COVID-19 diagnosis. Using X-rays, ML methods can be employed to identify COVID-19 patients by quantitively examining chest X-rays effectively. Therefore, we introduce an alternative, robust, and intelligent diagnostic tool for automatically detecting COVID-19 utilizing available resources from digital chest X-rays. Our technique is a hybrid framework that is based on the fusion of two techniques, Neutrosophic techniques (NTs) and ML. Classification features are extracted from X-ray images using morphological features (MFs) and principal component analysis (PCA). The ML networks were trained to classify the chest X-rays into two classes: positive (+ve) COVID-19 patients or normal subjects (or −ve). The experimental results are performed based on a sample from a collected comprehensive image dataset from several hospitals worldwide. The classification accuracy, precision, sensitivity, specificity and F1-score for the proposed scheme was 98.46%, 98.19%, 98.18%, 98.67%, and 98.17%. The experimental results also documented the high accuracy of the proposed pipeline compared to other literature techniques. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.