New chest x-ray tool uses AI to detect COVID-19 with 83% accuracy

Northwestern University researchers announced success on November 24th Developed A new radiation tool equipped with artificial intelligence (AI).

This tool is called DeepCOVID-XR and its AI algorithm can find COVID-19 from chest x-ray images about 1-6% more accurately and 10 times faster than a professional chest radiologist. Forbes Report..

(Photo: CDC / Unsplash)
The new chest x-ray tool uses artificial intelligence to detect Covid-19 with 83% accuracy

DeepCOVID-XR uses artificial intelligence to detect Covid-19

Evanston, Illinois-based university Journal Radiology, This shows that the new tool “classified 2,214 test images with 83% accuracy”. The authors also stated that early detection of coronavirus can protect healthcare professionals and patients by isolating patients earlier and flagging potential carriers of the virus.

To train and test machine learning tools, researchers used 17,002 chest x-ray images. Of these, 5,445 were from COVID-19-positive patients at various sites in the Northwestern Memorial Health System.

The team tested the Deep COVID-XR with 300 test images from Lake Forest Hospital. The researchers compared the results of the tool with those produced by five experienced cardiothoracic radiologists. Compared to a radiologist who spent about 2.5-3.5 hours analyzing these images, the DeepCOVID-XR AI system took only about 18 minutes to produce results.

However, Northwestern University AI expert and senior author of the study, Aggelos Katsaggelos, said he would not replace the actual COVID test. “X-rays are routine, safe, and cheap,” Katsaggelos said, adding that not only “determine if a patient needs to be quarantined”, but it only takes a few seconds to screen a person. It was.

(Photo: Go Nakamura / Getty Images)
Houston, Texas-November 26: (Editorial use only) Medical staff percutaneous tracheostomy in patients in COVID-19 Intensive Care Unit (ICU) during Thanksgiving at United Memorial Medical Center on November 26 Monitor the endoscope when performing open surgery. 2020 in Houston, Texas. Texas has reportedly reached more than 1.22 million cases, including more than 21,500 deaths.

Another author of the study, Dr. Ramsey Wehbe, explained that the ability to use algorithms to flag patients could significantly speed up the triage process. The COVID-19 test can give results in hours or days, greatly reducing the risk while you wait for the test results.

Katsaggelos said conducting initial screening for patients in need of quarantine can provide significant support to the emergency department. Healthcare workers and physicians were at increased risk of becoming infected with COVID-19 because they encountered COVID-19 patients seeking medical care.

However Press statement We warned that not all COVID-19 patients show signs of coronavirus on chest x-rays. This is especially true in the early stages of progression when the patient has not yet shown symptoms in the lungs. Therefore, the DeepCOVID-XR tool is not very useful in this case.

In fact, when a pandemic occurs, there is a shortage of personal protective equipment (PPE). This has led healthcare professionals to reuse PPE supplies and become infected with the virus.Originally Centers for Disease Control and Prevention (CDC)Nearly 238,000 health care workers have already been infected with the coronavirus, of which more than 841 have died of COVID-19.

Read again: The BioVyzr COVID-19 helmet promises full protection against coronavirus better than face masks

DeepCOVID-XR is not definitive

On the other hand, previous clinical studies have shown that chest images of COVID-19 pneumonia show distinct features, so the journal article also provides important consideration for Deep COVID-XR technology.

However, the tool also has some limitations. Many of the findings seen in the images are non-specific and overlap with other illnesses, especially the type of viral pneumonia. In addition, up to 56% of symptomatic patients show normal chest x-ray results in the early stages of the disease, and it is not appropriate to “exclude” the infection using diagnostic imaging.

Therefore, the authors wrote that chest images should not only be used as a COVID-19 diagnostic tool in their study. Instead, it may play an important role in speeding up the identification and isolation of patients who are “probably suffering from diseases that help with triage and infection control.”

“This algorithm may benefit healthcare systems that reduce unnecessary exposure to the virus by acting as an automated tool to quickly report suspicious chest images to patients for isolation and further examination. I feel The author made a conclusion..

Related article: Oxford University Develops “Very Rapid” COVID-19 Antigen Test Kit-Get Results in 5 Minutes!

It is owned by TechTimes

CJ Robles Screenplay Work

Ⓒ 2018 TECHTIMES.com All rights reserved. Please do not duplicate without permission.