The announcement came without much fanfare. In February, Goa's health minister floated the idea of making AI screening mandatory for every chest X-ray performed in the state's public hospitals. The proposal wasn't speculative posturing—Goa had just crossed 100,000 AI-assisted screenings in its lung cancer detection program, and the results were compelling enough to warrant a policy most places wouldn't dare attempt for another decade.
It's the sort of forced innovation you see when the alternative is systemic failure.
Consider the math. India has, at most, 1.9 IRIA-member radiologists per 100,000 people—an upper bound based on the roughly 27,700 members counted by the Indian Radiological & Imaging Association against a population closing in on 1.5 billion. This reflects IRIA membership, not necessarily all practicing radiologists. The actual number of radiologists available in the public system? Almost certainly lower. Meanwhile, the diagnostic imaging market keeps expanding—projected to hit somewhere near INR 21,000 crore, growing close to 10% annually. More machines, fewer people trained to read what they produce.
It's an uncomfortable asymmetry. And it's creating conditions where AI adoption isn't just encouraged—it's becoming inevitable.
When Scarcity Drives Speed
India's radiology gap doesn't exist in isolation. The country had roughly three MRI machines per million people as of 2019—the most recent consolidated baseline data available—and less than one radiation therapy unit per 10 million, according to OECD figures. Infrastructure is expanding, slowly. Expertise? Not so much.
Dr. Arjun Kalyanpur, whose Teleradiology Solutions has been navigating this terrain for years, draws a parallel to teleradiology itself—a technology that emerged precisely because the workforce couldn't keep pace with demand. AI, he's argued repeatedly, follows the same adoption arc: augmenting productivity and quality when human capacity hits a ceiling.
The numbers bear that out. India's teleradiology market reached INR 3,812.8 crore in 2025, growing at something like 23.7% annually. Companies like 5C Network claim to process upwards of 15,000 scans daily across 2,000-plus hospitals and diagnostic centers, supported by around 400 radiologists, with turnaround times averaging 30 minutes. That kind of throughput only works if AI is doing the first pass—triaging urgent cases, routing by acuity, queuing the routine stuff for batch review.
What's striking, though, is how small the dedicated AI radiology market remains. Fortune Business Insights pegs India's slice at about $21.6 million in 2026—barely 3.6% of the global total. The gap between infrastructure spending and AI adoption suggests this is early innings. The real question is how fast the market matures once procurement pathways and regulatory frameworks solidify.
Three Forces Converging
Three dynamics are pushing AI radiology forward in India, and they're starting to interlock.
First: explicit policy endorsement. The World Health Organization's March 2021 recommendation on computer-aided detection for tuberculosis screening via chest X-rays wasn't just symbolic. WHO shipped operational handbooks, calibration guidance—the works. Countries like India adopted it at scale. Qure.ai, one of the homegrown leaders in AI radiology, reported seven million TB screens in fiscal 2024-25 alone, deployed across 4,800-plus sites in 105 countries. That's not a pilot. That's infrastructure.
Second: cost-effectiveness data moving from theoretical to empirical. A rapid health technology assessment accepted in March this year by the Indian Institute of Public Health Gandhinagar—comparing Qure.ai's qXR and DeepTek's Genki platforms using 2023 data from Indian public facilities—found AI-assisted chest X-ray interpretation to be both clinically effective and cost-saving. The study landed in Frontiers in Digital Health in March. Procurement committees need that kind of evidence. One thing to claim savings; another to demonstrate them in actual Indian healthcare settings.
Third: regulatory clarity, arriving in stages. The Central Drugs Standard Control Organisation issued draft guidance in October 2025 on how Medical Device Rules apply to software-as-a-medical-device, including AI and machine learning applications. It's still draft form—always a caveat—but it signals a risk-based classification approach aligned with international frameworks (IMDRF, EU MDR). Separately, the Digital Personal Data Protection Rules were notified in November, operationalizing consent and breach protocols that affect health data processing directly. The Ayushman Bharat Digital Mission's integration requirements, updated as recently as February, mandate consent-based data sharing through standardized HL7 FHIR R4 protocols.
Taken together, these pieces reduce ambiguity. Companies now understand, more or less, what compliance looks like. That's progress, even if the rules aren't final.
Templates and Scale
Goa's lung cancer screening program, launched mid-2024 through a partnership between Qure.ai and AstraZeneca, offers a template for state-led adoption. By early this year, the program had screened more than 100,000 residents—a six-figure volume that demonstrates operational feasibility, even in a relatively small state.
Telangana followed suit, signing a memorandum of understanding in May to deploy Qure.ai's lung screening tech across public hospitals, citing rising lung cancer incidence over the past decade as justification. These aren't experiments anymore. They're production deployments with policy implications.
On the private side, 5C Network exemplifies the teleradiology model turbocharged by AI. The company's June blog post claimed 15,000-plus scans daily across 300-plus cities, with those 30-minute average turnaround times. You don't hit that throughput relying solely on human interpretation. AI flags urgent findings, routes cases, queues normals for batch review. Without it, the network of 400 radiologists wouldn't come close.
Then there's CARPL.ai, which represents something different: the orchestration layer. The company raised a $10 million Series A led by the International Finance Corporation in July (investment signed in May, deployed in June). CARPL's platform aggregates over 300 AI applications from 100-plus vendors, handling integration, validation, compliance work that individual hospitals struggle with. IFC's involvement signals confidence that the infrastructure for healthcare AI—not just the algorithms—is investable.
Qure.ai's trajectory is instructive on its own. The company closed a $65 million Series D in 2024 to expand tuberculosis, lung cancer, and stroke programs. By fiscal 2024-25: seven million TB screenings, 4,800 sites globally. Founder Prashant Warier told Forbes India last June that the company's focus is shifting healthcare from reactive to proactive—catching disease earlier, when intervention is cheaper and more effective. It's a pitch, sure. But one backed by deployment scale that's hard to dismiss.
DeepTek, while less prominent in mainstream coverage, earned WHO recommendation status for its Genki platform and was included alongside Qure.ai in that 2026 health technology assessment—positioning it as a credible alternative in the TB screening segment.
What Happens Next
The next 18 to 24 months will likely determine whether AI radiology scales across India or remains confined to pockets of excellence. A few things to watch.
First, finalization of CDSCO's software-as-a-medical-device guidance. The October 2025 draft provides a framework, but until it's formally adopted, vendors navigate gray zones. Post-market surveillance and clinical validation requirements—mentioned in CDSCO presentations as recently as July—will define how rigorously AI claims get tested after deployment.
Second, expansion of state-led screening programs beyond Goa and Telangana. If those pilots produce measurable reductions in late-stage cancer diagnoses or TB treatment delays, other states will follow. The Ayushman Bharat Digital Mission's consent infrastructure, updated in February, makes cross-facility referrals and data sharing technically feasible in ways they weren't before. That interoperability could unlock population-health surveillance at genuine scale.
Third, evolution of AI orchestration platforms like CARPL.ai. If enterprise buyers can deploy and monitor multiple AI tools through a single integrated layer—reducing procurement friction, ensuring ABDM compliance, simplifying audits—adoption curves steepen. The IFC's $10 million bet suggests international development finance sees this as plausible, not speculative.
Fourth, the evidence base. That rapid HTA published early this year is a start, but health technology assessments influence procurement budgets. If HTAIn—India's Health Technology Assessment body—continues evaluating AI imaging tools and finds consistent cost-effectiveness, central and state governments will have justification for large-scale buys.
There are risks, naturally. Nature ran coverage in March on radiology deepfakes capable of fooling radiologists—highlighting the need for secure pipelines and quality assurance baked into deployment, not tacked on afterward. AI systems are only as reliable as the data they're trained on and the environments they operate in. India's ICMR issued ethical guidelines for AI in healthcare back in 2023 addressing consent, bias, data quality, accountability. Guidelines don't enforce themselves, though.
The radiologist workforce will need to adapt, too. A June preprint debate on the future radiologist role through 2035 suggests the profession is still grappling with what "AI-augmented radiologist" actually means in practice. Some see efficiency gains. Others worry about deskilling. Dr. Kalyanpur's framing—that AI augments rather than replaces—sounds reassuring. Whether it holds up as volumes rise and AI handles more routine interpretation remains to be seen.
India's Production Linked Incentive Scheme for Medical Devices reported INR 1,153 crore in actual investment as of July, including radiology and imaging equipment. Domestic manufacturing of CT scanners, MRIs, and other devices is underway. If local OEMs bundle AI directly into hardware—as GE HealthCare's India-designed Revolution Aspire CT scanner does with AI reconstruction—the technology becomes harder to avoid. Buyers get AI whether they asked for it or not, and regulatory frameworks will need to account for that embedded intelligence.
The India AI Mission's health verticals, launched in February under the SAHI (Strategy for AI in Healthcare in India) and BODH (Benchmarking Open Data Platform for Health AI) frameworks, include Cancer AI CATCH grants providing up to INR 1 crore for scale-up. That's national policy explicitly funding AI healthcare innovation. Combined with the FDA authorization momentum—258 AI/ML medical devices cleared globally in 2025 alone, 76.6% in radiology—India is operating in an environment where the question isn't whether AI will be used, but how quickly and under what governance.
Bridging Time
The 100,000-plus scans in Goa aren't an endpoint. They're proof that AI radiology can operate at scale in resource-constrained settings. If cost-effectiveness holds and regulatory clarity arrives, the model will replicate.
India's healthcare gaps are too severe for incremental solutions. AI won't solve everything—overpromising would be irresponsible. But it might bridge the time until India trains another 50,000 radiologists, a timeline measured in decades, not years.
That's the uncomfortable truth driving adoption: when scarcity is acute enough, experimentation becomes necessity. Whether the radiologists, regulators, and patients are ready or not.
