We use cookies to keep our website secure, personalize your experience and for web analytics purposes. Read our Privacy Policy to learn more. By clicking Accept, you agree to our use of cookies.
Despite the remarkable scientific and medical breakthroughs that have been made with regards to COVID-19 remedies over the last 3 years, widely accessible treatment options for those afflicted with the Novel Coronavirus are still largely unavailable. Therefore, rapid and accurate detection of the virus via in-person screening has remained absolutely imperative to containing mass transmission within communities and throughout the globe.
The current gold standard for patient screening is referred to as reverse transcriptase-polymerase chain reaction, or RT-PCR, which essentially works by inferring the presence of COVID-19 based on the analysis of respiratory samples. While this method generally yields a high degree of result accuracy, PCR testing is a highly involved manual process with a relatively slow average turnaround time that can take up to several days after the test is performed for patients to receive results. Additionally, its lack of standardized reporting and its variable sensitivity make it a less-than-perfect method of screening. This is where the application of X-Ray imaging systems in conjunction with Artificial Intelligence (AI) and deep learning models can play a vital role in the detection of COVID-19.
In a Joint Research Paper published in ScienceDirect, Kanda highlights the results of a recent study conducted to evaluate the effectiveness of Indirect Supervision applied to the classification of COVID-19 and Pneumonia.
In this study, we demonstrated a training pipeline based on indirect supervision for neural networks, and after training a set of deep learning models, we found that the proposed pipeline allowed for a higher degree of accuracy with regards to the classification of COVID-19, as well as distinguishing its presence from that of Pneumonia. Based on the data collected from the performed experiments, the networks that were trained on the proposed pipeline performed comparably to practicing radiologists when it came to the classification of multiple thoracic pathologies in chest X-ray radiographs.
By implementing deep learning models to assist in an analytical capacity within the field of medicine, we can enable doctors and healthcare professionals to allocate more time and resources to high quality care and treatment, while simultaneously improving the accuracy and turnaround time of patient screenings. Read the full article to learn more about our machine learning training pipeline and its potential to improve healthcare delivery by expanding accessibility to chest radiology expertise for the detection of a variety of acute diseases.
