NEW DELHI – Reversing decades of negative messages, the World Health Organisation is once again endorsing the use of X-rays as a TB screening tool  in lower-income countries – this time in conjunction with the use of new artificial intelligence programmes that can read digital x-rays and identify suspected TB cases more accurately. 

For community-level screening of TB, the WHO has ranked the tools that could be used. The guideline says that if resources are available, the state should first use chest X-rays at the community level since an abnormal chest X-ray is very likely to be a TB-positive case. However, patients with abnormal chest X-rays have to undergo rapid molecular diagnostics to confirm TB. The other tools that can be used are deploying rapid molecular tests at community level, or screening of symptoms.  

The WHO guidelines also give a huge boost to computer-aided AI detection software and said that it can better interpret the digital X-rays, and triage suspected cases, more accurately than human X-ray readers. 

The new AI makes the screening process more efficient, and significantly reduces the time taken to make a diagnosis. The WHO has said that they would release a more detailed guideline this month, coinciding with World TB day, on Wednesday (March 24).

The move has been welcomed by TB experts as a tried-and-true means of screening – which was effectively used by countries for over a century, but falling by the wayside in recent decades. 

“This is huge,” said Salmaan Keshavjee Director, Harvard Medical School Center for Global Health Delivery. “It can find more people with potential disease. It was the approach that was used in Western countries since the early 20th century, so it’s more than 100 years old.”

In December last year, the WHO released the Rapid Communication on Systematic Screening for Tuberculosis recommended for community-wide screening, and particularly for HIV-positive people, pediatric contacts of TB patients, among others. The WHO released a consolidated guideline on systematic screening of TB recently which includes the use of chest X-ray and artificial intelligence. 

“This Rapid Communication is being issued to help national TB programmes and other stakeholders prepare for the changes that will be introduced with the new guidelines on TB screening.” the WHO said in its statement.

10 Million People Annually Diagnosed with TB – Mostly In Asia & Africa 

Globally, an estimated 10 million people fell ill with TB in 2019. Eight countries accounted for two-thirds of the global total including India, Indonesia, China, Philippines, Pakistan, Nigeria, Bangladesh and South Africa. India has the highest TB burden among these nations and accounts for 26% of the global total number of TB cases.

Many lower-income countries, including India still use smear microscopy as the first line of testing for most of the population. Smear microscopy involves simply looking for the bacteria through the microscope, a method which misses about half of TB-positive patients. The more sensitive diagnostic test- that is the Cartridge-based Nucleic Acid Amplification test or CB-NAAT — is mostly used for those who are sputum-positive to detect resistance to one TB drug- rifampicin. 

“We miss too many people with TB, and a relatively simple test like X-ray might help us find more people with TB. And AI-based software like Computer-aided detection software could help us do that, even when trained radiologists are not there for example in the rural or remote areas,” said Madhukar Pai from McGill University in Montreal, Canada.

Messaging On X-Ray Went Off The Tracks  

When TB programmes evolved, X-rays were a mainstay of diagnosis. Historically, miniature radiography for mass TB screening activities was widely used in high-income countries throughout the 20th century. 

In the early 90s, the WHO declared TB a global emergency. It advocated TB programmes to follow Directly Observed Treatment Short-course or DOTS. DOTS categorically recommended a limited use of X-ray, mostly as a supplemental diagnosis if sputum microscopy failed. 

This meant that patients who were not sputum-positive had to wait for chest X-rays to be ordered but continued to have TB symptoms, depending on who was treating them. 

“But the messaging was off the track. The idea was to say X-ray could be a diagnostic tool, but not a confirmatory tool. But somewhere along the line, doing X-ray was a sin, and only bad doctors use X-ray,” said Shibu Vijayan, Global TB Technical Director at PATH. The organisation engages the private sector in diagnosing and treating tuberculosis. 

The guidelines reflected a differential treatment for those in the lower and middle income countries and higher income countries, say some global health experts like Keshavjee. 

“DOTS did not recommend it  (use of X-ray) because they saw it as being too expensive.  It’s always been known that it is more effective. So they are correct to come back to it finally,” he said.

It was during the TB prevalence studies in countries such as Vietnam, Kenya and Zambia that found chest X-rays were detecting more patients, as compared to sputum microscopy.

 “It was found that around 40% of those surveyed with X-ray shadows and were positive for TB did not have any symptoms at all. That’s how experts asked for X-ray to be brought back,” said Vijayan. 

The focus is now, however, on using X-ray as a screening tool, and not a diagnostic tool, per se.

“We now know that X-rays are good for triage or screening, to find out who needs further confirmatory testing. So, someone with TB symptoms could get an X-ray, and if that is abnormal, then CBNAAT could be done. So, the current focus is really on screening, not diagnosis. For diagnosis, WHO still endorses rapid molecular tests,” said Pai.

Lower Cost Digital X-Rays & AI Been Game Changers

The game changer has been the relatively low cost of digital X-rays and artificial intelligence programmes that enhance the efficiency of the screening process.

Friends for International TB Relief (FIT) in Vietnam has been conducting chest X-rays screenings across the country, even in remote parts. It works with multiple global agencies focusing on implementing TB and HIV programmes in Vietnam. 

“I think my organisation in Vietnam has shown that chest X-ray screenings can be done anywhere,” said Andrew Codlin from FIT. 

“The new portable X-ray machines are the size of digital cameras, something that you can put in the backpack and walk up the hill.  We have done screening campaigns in a remote island, in mountainous areas. We also had a screening camp during a cyclone with water running through the streets,” he said.

“If there is commitment, and the right buy-in from the political establishment, it is not difficult to scale up screening campaigns,” Codlin said.

Qure.ai  is a computer-aided technology that can detect abnormalities in the chest X-ray and is used in 20 countries. It is particularly useful in mass screening camps.

“Our product -qXR processes the X-rays that are recorded in a cloud. The health worker can see the report within a minute of taking the X-ray in an app,” said Prashant Warrier, founder of Qure.ai. In case of an abnormal X-ray, the patient’s sputum sample can then be sent for rapid molecular diagnostics.

In Nagpur, a city in West India, PATH helped run a pilot programme with a local nonprofit, Disha Foundation using chest X-ray with a computer-aided detection software called Qurei in the private sector. 

In India, more than half of the TB patients are treated in the private sector. Presumptive TB patients were provided a free X-ray under this pilot. If the patient had an abnormal X-ray, the sputum samples would be sent for rapid molecular testing in a medical college there. 

“When patients do not feel better, they switch doctors. The quick turnaround of the AI technology helps retain patients in the programme and ensure they take treatment,” said Lucky Richardson Masih, operations manager, Disha Foundation in Nagpur. 

The organisation works with local private doctors in Nagpur’s slums in providing the patients with approved diagnostic facilities and treatment. They tied up with PATH for this project in 2019.

The use of radiological screening and AI resulted in a 13% additional TB cases being detected, said Vaishnavi Jondhale, Operations Manager, Path Mumbai.

Warrier claimed that the product can find more cases, and is far more sensitive than a radiologist reading X-ray reports. The cost works out cheaper as well.

Codlin explained how AI is more sensitive. 

 “If a human reader reports 100 abnormal chest X-rays, perhaps 10 would be positive for TB.  If we use qXR it will report 50 abnormal chest X-rays for 10 TB positive cases will be found. We are effectively using fewer CBNAAT tests and resources that way,” he said.

 

 

 

 

Image Credits: Andrew Codlin.