How portable, AI X-ray machines are helping Uganda beat TB

The road to Bukibokolo village in the hills of Bududa, eastern Uganda, was steep and rough – impassable on two wheels, Moses Eyaru concluded. But the health worker needed to screen the village dwellers for tuberculosis (TB) and that couldn’t wait.

Ministry of Health data had shown that Bukibokolo was one of the many TB hotspots in the east African country. So, Eyaru got off his motorcycle and hauled a backpack, which contained a digital X-ray system, onto his back and, together with his team-mates, slowly climbed the rough road towards the village.

“This is Uganda – our roads are not always the best. But these digital X-ray systems that are equipped with computer-aided detection (CAD) or artificial intelligence (AI) for screening TB are helping us to reach some of these hard-to-reach areas,” said Eyaru, who is a radiographer at Bududa Hospital. “The tools are portable and solar-powered. The backpack weighs about 35 kg and can be transported by car, boat or motorcycle.”

“Ordinarily you require an expert or a radiologist to interpret an X-ray image, but with artificial intelligence, the programme will interpret the image.”

– Dr Deus Lukoye, TB expert, US Centers for Disease Control and Prevention, Uganda

When Eyaru got to the village, after about 20 minutes, he explained to a group of people gathered in the village’s community hall that he had come to screen for TB. He explained the common TB symptoms, such as cough, fever, weight loss and poor appetite, but most importantly he emphasised that the disease was treatable when detected early. After this, he unzipped the backpack and pulled out a laptop, the X-ray system, developed by Dutch company Delft Imaging specifically for use during mobile outreach, and other accessories.

He assembled the system, put on a protective apron and started to screen the villagers one after the other.

“The system is automated and does not require special health workers to interpret chest images,” said Eyaru in an interview with VaccinesWork. “I just connect the digital X-ray to the CAD4TB software or AI and that is it. After the X-ray has taken the image, it feeds the information to the CAD, which interprets it and relays real-time info indicating that TB is likely or not.

“After this, we use a GeneXpert machine [a molecular testing device] to diagnose those whose results are suggestive of TB,” he said. “The GeneXpert confirms if they have TB or not. Those who test positive for TB will be referred to hospital or nearby health facilities and started on treatment.”

“Digital X-rays that are equipped with AI are helping us to actively screen and diagnose TB in key hard-to-reach areas that may not have existing health centres,” said Dr Aldo Burua, senior TB case-finding advisor at the national TB and leprosy programme, and coordinator for X-ray services in Uganda. “The tools are fast and take seconds to capture clear images. And because we are using AI to interpret the images, we can screen more than 150 people in a day.”

Since 2020, Uganda has invested in 22 digital X-ray systems, in five mobile vans and 17 backpack-kits, that are helping extend TB screening to hard-to-reach areas across the country.

The country has also invested in about 300 GeneXpert machines, which are installed at its health centre IV’s and hospitals. These interventions have helped check an additional 50,000 people for TB, diagnosing almost 1,000 new patients with the bacterial infection between 2022 and 2023, Burua said.

The interventions have helped to improve TB case notifications in Uganda. As of today, the Ministry of Health screens and tests almost 90% of the estimated 94,000 people who contract TB in the east African country every year, compared with the 60% it screened and tested five years ago.

Catch and kill

As in much of Africa, TB is a leading cause of death in Uganda, killing about 15,600 people annually. “The challenge is that we have a very poor health-seeking behaviour,” said Burua. “People do not report to health facilities until they are really sick, when the disease has affected their lungs. They stay in communities spreading the bacteria.” But digital X-ray systems that are equipped with AI, coupled with other modern testing tools such as the molecular GeneXpert, offer a means to catch infections early.

“When we start people on treatment early, they are more likely to respond to the medicine and not to spread the disease,” said Burua. “From our data, we know where these TB cases are. So, we organise outreaches to these specific hotspots.”

Initially, health officers send village health teams to mobilise the communities, enlisting the help of local media, churches, and local council workers, in order to identify areas that need screening. After that, health workers such as Moses Eyaru bring in the X-ray systems.

Typically, the screening team includes clinicians, radiographers, TB nurses and lab technicians. They use a GeneXpert machine to diagnose those who are suggestive of TB. If they don’t have a GeneXpert machine to confirm TB, the health workers collect sputum samples from persons who are suspected to be infected, take their details and contact them when results are out.

Triaging the vulnerable

“We target people who are at high risk for contracting TB,” said Burua. “People who are living in homes which have TB patients and those with other comorbidities such as diabetes or are malnourished, or living with HIV.” The team also targets people who live in congested places, such as refugees, prisoners and slum dwellers.

Previously, health workers based at facilities relied on verbal histories of symptoms related by patients, followed up by a request for a sputum sample in case of suspicion. After that, a laboratory technician would smear that sputum onto a glass slide, and peer at the slide under a microscope to establish a diagnosis.

Dr Deus Lukoye, a TB expert at the US Centers for Disease Control and Prevention in Uganda hails digital X-rays and intelligent computer programmes as “advancements”. The programmes read X-ray images in real-time, recognising telltale patterns of abnormality, he explains.

“Ordinarily you require an expert or a radiologist to interpret an X-ray image, but with artificial intelligence, the programme will interpret the image,” said Lukoye.  

The actual 35kg backpack aside, new technologies are easing the burden carried by radiologists like Eyaru. Higher image quality makes seeking second opinions from distant experts easier, and can help minimise waste of GeneXpert cartridges, because these are now only deployed to diagnose cases that the X-ray tech has signalled as suggestive of TB, Eyaru said.

“With the microscope [alone], health workers would be exhausted after examining about 10 patients,” he said. With the new tools and the automated GeneXpert, by contrast, a health worker can screen more than 150 people and diagnose up to 30 people in a day.

Ideally, health workers need a diagnostic test to confirm TB before they can start people on treatment, said Burua. However, for various reasons, including a low TB load in a patient’s sputum, TB germs may not show up on available diagnostic tests, and health workers have to diagnose the presence of disease from symptoms. 

In cases like these, a health worker can start the patient on treatment if their X-ray and its AI-generated read-out is suggestive of TB, and they are showing symptoms. “The digital X-ray and AI are able to recognise abnormalities that health workers may not pick and have also helped to diagnose latent TB [in which the body contains the germ, but it’s dormant and may not be actively inducing symptoms],” Burua said.

The GeneXpert machine can also pick up whether the present strain of TB is the multidrug-resistant TB (MDR-TB), enabling health workers to start patients on appropriate treatment. MDR-TB affects about 1% of new TB cases in Uganda.

Burua said the ministry was lobbying to have a digital X-ray machine in each of Uganda’s 135 districts.