2 days ago
India's TB fight now has an X factor: AI-powered portable kit for early, fast detection
At the Villianur Health and Wellness Centre in Puducherry, the family member of a tuberculosis (TB) patient steps in for a quick X-ray and gets in front of a plate, lifting up his necklace over his head to prevent disruption. A lead apron-clad technician holds up a portable device that looks like an SLR camera with a black lens and clicks an image. Within minutes, the result is out — he is declared TB-free.
In that short span of time, the images of the chest X-ray had shown up on the laptop at the TB screening desk where an AI tool, fed with chest scan data of TB patients, had found no white patches in his lungs.
This hand-held, AI-enabled X-ray machine has been recently added as a screening tool in the national TB programme. Such quick screening helps identify undiagnosed TB cases within the patient's family, enabling early treatment and preventing further transmission within the household and community, particularly in remote areas.
How TB screening goes the last mile
At present, there are 473 such hand-held devices in use across the country. The government is in the process of procuring another 1,500 to diagnose missing TB cases. It is being used for screening vulnerable individuals — such as family members of TB patients, those with respiratory symptoms and those living with diabetes among others. It is also used for finding active cases in places where camps are conducted to test the already mapped vulnerable population in a district.
'Anyone who is found to be positive on the X-ray is asked to submit a sputum sample. They are given the sample container, asked to step out in the open area, and cough up sputum for testing. The sample is then sent to a government laboratory for testing,' said Dr Pamagal Kavithai, medical officer in-charge of the health centre.
A mobile device helps in remote pockets
While the devices work on the same principle as the stationary X-ray machines found in hospitals, they are highly portable and battery-operated, making them ideal for use in hard-to-reach areas with limited resources. The devices are extremely light — weighing between 1.8 kg and 2.8kg — and can be carried around in a backpack.
The device comes as a set of a digital SLR like X-ray beam generator, a flat digital panel X-ray detector, tripod to mount the detector and a laptop with in-built AI for reading the X-rays. The kits also contain a lead apron for technicians. The devices can take around 100 X-rays with a single charge — useful for camps in remote areas.
What are ICMR-validated devices?
There are three such AI-enabled X-ray devices that have been validated by the Indian Council of Medical Research (ICMR) — LabIndia's Mine2, Mylab's MyBeam, and Prognosys' Prorad Atlas. In fact, Prognosys is a sister company to Molbio, which is behind the cost-effective, portable, battery-operated molecular diagnostic machine called TrueNat that is being extensively used in the country's TB programme. The validations have shown that these devices produce X-rays comparable in quality to the ones produced by digital X-ray devices found in hospitals.
What about costs?
'Over the years, the sensitivity of the X-ray detectors has improved, so has the collimation — meaning there is minimal scattering and leakage from the X-ray beam in the generator. Very little radiation is now needed to get good quality scans. This is the reason now we can have hand-held devices that do not require much infrastructure,' said Santosh Kumar, business head, LabIndia, which used to import the devices from a South Korean company and has now started manufacturing it in India through licensing.
'Currently, there are around eight to 10 companies in India that have started building such machines. Mylab was one of the first. This boom has happened over the last two years. Earlier, such devices would have to be imported from other countries and would cost Rs 40-45 lakh, with an additional Rs 5 lakh to Rs 7 lakh a year for the AI. Now, the costs have reduced to a third,' said Hasmukh Rawal, managing director of Mylab, which also makes RT-PCR kits for TB diagnosis.
'It is an ultraportable, battery-operated X-ray machine. It can charge on any wall socket for four to six hours. With that charge you can do over a 100 X-rays over 8-10 hours, which is pretty much the entire day. The need for an X-ray room and the constraint of bringing people to a hospital or clinic set-up is completely removed. You can get diagnostic quality X-rays within a few seconds,' said Dr Krishna Prasad, head of Prognosys.
'Chest X-rays are good for screening of TB cases early on before people start showing symptoms. If these cases are detected in the community — especially as they are unlikely to come to a health centre for testing otherwise — we can prevent further spread of the infection and eliminate the disease,' said Kumar.
'The government has also developed an AI tool which can be utilised by Indian companies. We work with a few AI providers, including an AI based on the ICMR database of X-rays,' said Rawal.
ICMR's open dataset
The ICMR consortium working on developing and validating these innovations for TB has also created an open dataset of chest X-rays that can be used by companies for developing such AI tools. Over the last few years, ICMR's National Institute of Research in Tuberculosis has been building a database for such AI applications.
With an archive of thousands of X-rays, scientists from the institute started training an AI model to first recognise abnormalities on an X-ray, then to distinguish between different types of abnormalities such as fluid in chest or cancerous tumours, and finally to definitively identify whether a person has TB.
These portable X-ray machines were introduced in the national programme during the 100-day campaign that started in December last year, where vulnerable people from 347 high-risk districts were tested to find TB cases in the community, especially those without any symptoms and who were unlikely to reach a health centre for TB testing otherwise.
How screening helped in drawing an accurate TB map
During the 100-day campaign, 12.97 crore people were screened, of whom 7.19 lakh people were found to have TB. Importantly, 2.85 lakh of these patients were asymptomatic.
There is usually a gap between the number of estimated TB cases in a country and the actual number of cases diagnosed — the difference is considered to be the missing cases, who may continue to spread the infection in the country. In India, this gap has been reducing, with more and more people getting diagnosed and receiving treatment.
There were an estimated 28 lakh TB cases in India in 2023, with 25.2 lakh actually getting diagnosed, according to the latest available Global TB Report. To compare, there were an estimated 28.2 lakh TB cases in 2022 and 24.2 lakh were diagnosed.
