The current device used for detection of the COVID-19 virus in humanbeings is based on a method called RT-PCR (Reverse Transcription-Polymerase Chain Reaction). While this was a revolutionary discovery, many errors are involved with this device to detect COVID-19. Results of possible infection are also time consuming, taking up to two days to get results.
In this respect, Satyavratan Govindarajan, a PhD scholar, and Professor Ramakrishnan Swaminathan have developed an easy-to-use new technique for immediate detection of COVID-19 in the body. They have come up with a proposal to use artificial intelligence for screening patients. For this method, Chest Radiographic (CXR) Imaging, coupled with simplified Convolutional Neural Network (CNN) is done along with occlusion sensitivity maps. CXR is similar to X-Ray, except that it is directed on the chest. Convolutional Neural Networks (CNNs) are advanced machine learning tools, when combined with CXR, provide representation of the image with essential lung characteristics. Occlusion is a visualization technique, when combined with CNN and CXR, offers clearer view of the area where the disease is present in the patient’s lungs. It helps to assist doctors to locate the area of the disease and explains to assess its severity. The two biomedical engineers strongly believe that such kind of a proposal would also enable even the inexperienced medical personnel to screen affected patients in any remote and resource poor regions.
The researchers claim that the proposed model would give faster results, more clarity and clearer images of the chest where infection occurs. One hopes that future work would involve full-scale research on implementing deeper architectures of CNN models using larger image datasets and exploring the need for lung field segmentation for better detection of the disease.
Prof.Ramakrishnan Swaminathan 
Satyavratan Govindarajan
Dr. Alladi Mohan, MD (Medicine), (AIIMS, New Delhi) made the following comments on the work of the two engineers: “A chest radiograph tells us “where” the disease is; but, NOT “what” the disease is. This is true not only for COVID-19 but for any chest disease as well. Several “viral” pneumonias, including COVID-19 would look the same on CXR. Even though imaging categorization has been postulated as an “index of suspicion” (like CORADS system on CT of the chest), definitive diagnosis of COVID-19 in real life clinical setting would be by doing the nasopharyngeal swab RT-PCR testing. Nevertheless, automated differentiation of COVID-19 patients from healthy subjects using chest X-ray images as has been done in this paper is a good academic exercise. The practical utility and application of this automated differentiation model’s discriminatory function in real life setting merits validation.”
Article by Akshay Anantharaman
Here is the link to the research article:
https://link.springer.com/article/10.1007/s10489-020-01941-8
