A Japanese medical AI company has begun deploying diagnostic imaging tools across Vietnam's nationwide healthcare network, marking one of the first large-scale regional rollouts of AI-assisted diagnostics in Southeast Asia, even as U.S. regulators grapple with how to safely govern these same technologies. LPIXEL Inc. has partnered with VNPT Information Technology Company (VNPT-IT), the IT subsidiary of Vietnam's largest telecommunications operator, to integrate its "EIRL" medical imaging AI series into cloud-based systems serving over 600 healthcare institutions across the country.

Why Is ASEAN Suddenly Embracing Medical AI?

Southeast Asia faces a healthcare crisis that AI is uniquely positioned to address. The region is experiencing a dramatic shift in disease patterns, with non-communicable diseases like cancer, cardiovascular disease, and diabetes now accounting for approximately 62% of all deaths, up from infectious diseases that once dominated. At the same time, the region faces a severe shortage of medical professionals. Vietnam has only 0.83 physicians per 1,000 people, compared to Japan's 2.61 and Indonesia's 0.70. This gap means radiologists are overwhelmed with diagnostic imaging workloads, creating urgent demand for AI-assisted diagnostic support.

The partnership between LPIXEL and VNPT-IT directly addresses this bottleneck. EIRL's AI-based analysis of medical images, particularly chest X-rays through its "EIRL Chest Screening" tool, integrates seamlessly into existing hospital workflows. When a radiologist orders an image analysis, the system automatically sends the image to EIRL via VNPT's DICOM Gateway (a standard medical imaging protocol), receives AI-generated analysis results, and displays them directly in the radiologist's existing imaging software.

What Does Successful Deployment Look Like in Practice?

The partnership builds on a proof-of-concept project launched in September 2025 at more than 10 Vietnamese healthcare institutions, including Hanoi Post Hospital. Following the trial period, radiologists reported strong clinical value and rapid adoption. Dr. An, Head of Radiology at Hanoi Post Hospital, noted that EIRL Chest Screening "provided clear clinical value and quickly gained the trust of frontline physicians". Based on this validation, LPIXEL and VNPT-IT have now moved to formal commercial deployment, with full system implementation already completed at more than 10 healthcare institutions across Vietnam.

The success of this model reflects a broader shift in how medical AI is being deployed globally. Rather than selling individual AI tools to hospitals, companies are now integrating AI into existing healthcare IT infrastructure, making adoption faster and less disruptive to clinical workflows.

How Is LPIXEL Expanding Across ASEAN?

• Regulatory Approvals: EIRL has obtained medical device regulatory approvals in six ASEAN markets: Vietnam, Thailand, Indonesia, the Philippines, Malaysia, and Singapore, removing a major barrier to regional expansion.
• Partnership Strategy: LPIXEL is actively seeking distribution partners in Thailand, Indonesia, and the Philippines that have strong local healthcare networks and expertise in medical IT infrastructure, including PACS/RIS (Picture Archiving and Communication System/Radiology Information System) systems.
• Infrastructure Leverage: By partnering with VNPT-IT, LPIXEL gains access to Vietnam's nationwide digital healthcare network, which serves as a model for how AI can be deployed at scale across developing healthcare systems.

What Regulatory Challenges Does AI in Healthcare Still Face?

While LPIXEL's deployment represents progress in medical imaging diagnostics, regulatory uncertainty remains a significant challenge for AI healthcare tools globally. The Association for Diagnostics and Laboratory Medicine (ADLM) recently submitted formal comments to U.S. federal regulators calling for updates to the Clinical Laboratory Improvement Amendments (CLIA), the regulatory framework governing all clinical laboratory testing in the United States. The organization highlighted that current regulations, written in 1992, do not adequately address the unique risks posed by AI and machine learning tools in laboratory medicine.

ADLM identified several critical gaps in the current regulatory framework for laboratory testing environments. Traditional software errors typically affect every patient case meeting the same programmed conditions and can be identified using test cases with known expected outputs. AI-based models, by contrast, may produce case-specific errors that are far more difficult to troubleshoot. Additionally, generative AI systems can produce inaccurate or unsupported information, omit clinically important facts, or change behavior following updates to the underlying model or knowledge base.

"AI has the potential to support tremendous advances in laboratory medicine, but innovation in this area must be balanced with the need to ensure test quality and patient safety," said Dr. Stanley F. Lo.

Dr. Stanley F. Lo, President, Association for Diagnostics and Laboratory Medicine

Steps Regulators Should Take to Ensure AI Safety in Laboratory Medicine

• Establish Clear Oversight: The Centers for Medicare and Medicaid Services and the Centers for Disease Control and Prevention should establish federal oversight for AI-based tools that avoids unnecessary duplication with FDA regulation, ensuring that both the software product and the laboratory's use of it are appropriately governed.
• Distinguish AI from Conventional Software: CLIA should explicitly distinguish between conventional software and AI models, requiring laboratories to account for differences in how traditional software and learned models perform when validating and monitoring their performance.
• Adopt Risk-Based Standards: Regulators should establish a risk-based, technology-appropriate approach that sets clear quality expectations for AI-based tools while preserving the ability of laboratory directors and qualified professionals to determine scientifically appropriate methods for meeting those expectations.
• Extend Oversight to Interpretation Services: When a facility independently analyzes patient-specific laboratory data or provides specialized laboratory interpretation that generates or helps generate a test result for clinical use, those activities should be subject to appropriate CLIA oversight as part of the total testing process.

ADLM emphasized that AI-based tools should be evaluated as part of the total testing process within CLIA's existing framework rather than treated as standalone software tools requiring a separate regulatory structure. This approach mirrors the strategy LPIXEL is taking in Vietnam, where EIRL is integrated into existing hospital systems rather than deployed as an isolated tool.

Why the Speed Difference Between Regions Matters

The contrast between LPIXEL's rapid deployment in Vietnam and the ongoing regulatory deliberations in the United States highlights a fundamental tension in global healthcare AI adoption. Regions with less established regulatory frameworks can move faster, while mature healthcare systems like the U.S. must balance innovation with patient safety through more rigorous oversight. Vietnam's ability to deploy EIRL across 600 healthcare institutions reflects both the urgency of the physician shortage and the flexibility of emerging healthcare systems to adopt new technologies quickly.

However, this speed advantage comes with risks. Without robust post-deployment monitoring and validation frameworks, AI tools deployed at scale may encounter unforeseen failure modes that only emerge in real-world clinical settings with diverse patient populations. As AI diagnostic tools continue to proliferate across ASEAN and beyond, the regulatory frameworks governing their use will likely determine whether these technologies reduce healthcare disparities or widen them further.