A new study published in Nature Communications showed that an artificial intelligence algorithm can be designed to identify Covid-19 related pneumonia in computed tomography (CT) scans.

Researchers mentioned in the study that this could yield 99 per cent accurate results and can overcome the challenges of the present testing.

Researchers claimed that their AI algorithm correctly identifies positive cases 84 per cent of the time and negative cases 93 per cent of the time.

Study author Ulas Bagci from the University of Central Florida in the US said in a statement: “We demonstrated that a deep learning-based AI approach can serve as a standardised and objective tool to assist healthcare systems as well as patients.”

“It can be used as a complementary test tool in very specific limited populations, and it can be used rapidly and at large scale in the unfortunate event of a recurrent outbreak,” Bagci added.

Better than RT-PCR tests

The researchers believe that CT scans are a more effective tool for Covid diagnosis than the widely used reverse transcription-polymerase chain reaction, or RT-PCR, tests.

Researchers further said that RT-PCR tests have shown high false-negative rates, delays in processing, among other issues.

Also read: Ventilation systems could expose people to Covid-19: Study

In the study, researchers noted that CT scans are also effective in detecting asymptomatic cases of the virus. However, these were not recommended for Covid-19 as the infection seems similar to influenza-associated pneumonia on the scans.

But this problem can be overcome with the introduction of a new algorithm in CT scans so as to distinguish coronavirus and influenza, researchers said.

For the study, the researchers trained a computer algorithm to recognise coronavirus in lung CT scans of 1,280 multinational patients from China, Japan, and Italy.

The researchers then tested the algorithm on CT scans of 1,337 patients with lung diseases ranging from Covid-19 to cancer and non-Covid pneumonia.

They concluded after their tests that the algorithm was extremely proficient in accurately diagnosing Covid-19 pneumonia in the lungs while distinguishing it from influenza.

“We showed that robust AI models can achieve up to 90 per cent accuracy in independent test populations, maintain high specificity in non-Covid-19 related pneumonia, and demonstrate sufficient generalizability to unseen patient populations and centers,” Bagci said.