How Artificial Intelligence in radiology is pushing the boundaries of healthcare delivery and in public health management
September 27: TB is a disease which has been present for centuries. The TB burden is primary in SE Asia and also in India. WHO has set a goal for TB elimination by 2035, and India also set an ambitious goal to eliminate TB in the country by 2025. The focus is now on how healthcare organizations, individual practitioners and experts work together in both the public and private sectors to achieve these goals. TB detection to treatment is a team activity, and there are protocols in place which have evolved over time. Newer cutting-edge technology like Artificial Intelligence (AI) has been developed in the last few years, and gradually they have found adoption in diagnostic care pathways ensuring faster diagnosis and treatment and also ensuring the limit in the spread of the disease due to the nature of the intervention.
At DeepTek, we are working closely to create unique solutions to help strengthen the fight against TB. DeepTek is a Pune-based AI company which is developing AI solutions for radiology and imaging with a team of 120 members. At DeepTek, one of the key solutions we have been developing is the detection of TB from an x-ray chest. The AI software can also detect 20 other incidental or co-existing pathologies using AI.
How does it work in a real-world scenario:
To ensure faster elimination of TB, there has been a lot of impetus by the government to go to the vulnerable population group in high-risk populations by getting the screening and detection machinery and tools to them rather than waiting for the patients or suspected patients to come to the hospital.
Mobile health check-up units and vans are now commonplace, and they are equipped with a digital x-ray unit, sputum collection centre and patient registration portal for rapid detection. Recently we have also seen an upsurge of mobile diagnostic x-ray devices, which are handheld and the size of a DSLR professional camera and can do an x-ray of the chest at a low dose and of appreciable quality. These x-ray images can now be directly sent by wifi to a laptop.
The AI software created by DeepTek called “Genki” (Japanese word for healthy) resides on the laptop, and it does a handshake with the mobile x-ray unit using AE title, port and IP as all our home devices connect with each other. The system is so easy that it can be done remotely without anyone actually visiting the centre. This solution is called a Genki Edge Solution since it works on the local device.
Once the x-ray is sent to the software, it gets instantly analyzed by the AI within a 30-second window using the memory and compute capacity of the laptop to provide an instant result. In the result section, the AI software actually highlights the region of interest where TB is suspected and draws an outline view or bounding box around it. This allows the healthcare operator to understand where exactly the AI feels the pathology. This is all done offline without access to the internet (offline), allowing point-of-care diagnosis. This allows the operator to immediately make a note of the same in the Radiology Patient Registration Portal, which can be operated on mobile (again offline), and take the next steps, such as sputum sample collection. A simple intervention such as this can create a significant impact to ensure that the patient is not lost for follow-up, which is actually the weakest link in the patient diagnosis pathway.
The Genki Edge Solution in the van can now send this data to the Genki cloud solution, where it can be further processed by a host of 20+ AI models and showcase additional results besides the initial screening done on-site. Also, all abnormal studies can now be reviewed by expert radiologists/physicians to make a diagnosis and give an expert-validated report. This cloud solution also allows the portal to be accessed by nodal officers managing the TB screening program to ensure if vans are active, in the field, the number of x-rays done, AI results, radiologists report, report turnaround time and also diseases other than TB detected in the population and plan appropriate referral mechanism. Now they can also do a follow-up action plan and see progress monitoring of the disease. The cloud platform from DeepTek is called Genki Hub. It can be further synced to the Nikshay database.
The Genki AI solution for TB screening has been deployed in Greater Chennai Corporation for more than two years and has screened more than 200,000 patients. Currently, we are exploring this for statewide implementation. DeepTek solution is also being used at radiology clinics through a project across four northern and central Indian states through a project led by World Health Partners and USAID support. A few pilots of 2000 patients have also started in the van being operated out of Aundh Chest Pune and a medical college in Karnataka. Strategic tie-up and concept deployments have started in partnership with Dell in Mongolia and local partners in the Philippines and Kenya. Genki solution will also be onboarded onto portable x-ray machines of a global x-ray manufacturer Shimadzu for certain geographies across the Asia Pacific.
The system has been validated by independent testing and audits through STOP-TB WHO, Friends for International TB Relief, Vietnam. The results are available in Nature Scientific Reports as an independent audit of AI solutions and a peer-reviewed publication. It is documented that Genki Ai solutions have matched expert radiologists’ performance, and it has helped increase the yield of TB, ensuring better on-field screening.
The end-to-end solution can work not only for portable x-ray machines but also for hospital-based Computed and Digital Radiography systems and wherever there is a need for screening, triage and a radiologist opinion necessary. Currently, this initiative is done jointly with support from CSR partners like NTT DATA and implementation partners like Clinton Health Access Initiatives, World Health Partners, and USAID. We are happy to collaborate and provide this solution as and where needed through other organizations which are working in the social and health impact sectors and other CSR partners.
Moving beyond chest X-ray AI for Tuberculosis screening, DeepTek also has in its portfolio X-Ray AI for Knee and MRI Ai solutions for Spine to automate reporting and measurement (quantification process). DeepTek has created an industry-first AI platform for radiology image archival and distribution. This is now deployed across 400 plus centres across the country and Asia Pacific Region. The key feature of the solution named “Augmento” (Augemento is designed for Augmenting imaging experts) makes it easy for radiologists to report ensuring minimal burnouts amongst experts. This platform can be highly beneficial to diagnostic chains and hospitals as an AI PACS. The solution is available as a subscription-based service. This is India’s first AI Cloud PACS with a free Dicom viewer.
The radiological transformation from conventional imaging to digital like computed and digital radiology spurred a revolution and created extremely large opportunities for healthcare delivery. The initial apprehensions were laid to rest by the digital healthcare boom. Today technology like AI, smart workflow optimization and miniaturization of imaging equipment is fuelling the second level of exploration in radiology. This may lead to the consumerization of certain radiology services and may result in further penetration of radiology to the grass root level allowing ease of access. This will again create further job opportunities in the health care sector but may put unrealistic expectations on report turnaround time for radiologists. Modern Technological solutions like AI and workflow optimization AI platforms are designed with this future in mind to ensure robust healthcare delivery mechanisms and maximize reach.