Engg student claims COVID-19 can be detected by AI software
   Date :22-Jun-2020

Vedant Khairnar_1 &n
Vedant Khairnar
 
 
 
Principal Correspondent ;
 
Vedant Khairnar, Coder and student of Shri Ramdeobaba College of Engineering and Management (RCOEM), came forward in the fight of the nation with COVID19 by customising a software, which can be easily used to detect the COVID positive patients. The software is tested through the chest X-rays of covid patients, which are made available online by open sources.
 
The dataset of x-ray images is taken from, Kaggle.com, which is open source and publicly available, informed Vedant. The COVID19 patients chest-X-ray is different from normal and healthy persons chest-X-ray. This principle is used to differentiate the two persons as COVID positive and NORMAL. The job of differentiating these two can be done by the 'software program', called as classifier program. Initially the computer software is trained by giving it a set of large number of chest-X-rays images of covid infection. Once it is trained, the x ray image which we want to test is given to the software, and run command is given. The computer gives the result of detection of either positive or normal within a second.
 
This way the time and cost of detecting COVID-positive can be saved considerably. Though the model is a working one, it needs to be tested for more flexible dataset samples for our geographical locations, to bring this to actual implementation, informed Vedant The idea of developing this program is given by his father, Dr Khairnar who is professor at Government college of Engineering Nagpur. "Vedant was working on another project. It was regarding lung infections classifier software, and I thought, COVID is also a type of lung infections and I told him that the program on which he is working can also work for it, and result is in front of everybody," said Dr Khairnar.
 
"We have written to different medical authorities about our findings, and requested them to provide us, X-ray images of COVID patients and we can test them on our model for making the trial to bring it to actual use. Once the trial is over, it can then be used by government medical team for COVID detection. We are expecting a reply from them so that this can be put in actual practice. The accuracy of the application claimed to be 97%. This we want to do as a part of social service, we are going to make it 'Open source' shortly" said Vedant Khairnar.