AI in POCUS: An Overview of the Current Billing and Reimbursement

Joshua Guttman, MD, FAEMUS
Emory University School of Medicine

As Artificial Intelligence (AI) becomes increasingly integrated into medicine, reimbursement for AI-enabled tools has become an important and evolving conversation. Point-of-care ultrasound (POCUS) is no exception. Both third-party AI software and AI embedded directly within ultrasound devices are becoming more common and are being used with increasing frequency in clinical practice.

For readers who want a concise answer, the bottom line is summarized first, followed by background and detailed discussion.

The Bottom Line

POCUS that incorporates AI is reimbursed in the same manner as POCUS performed without AI, provided that the physician independently reviews and interprets the study. Documentation must clearly reflect that the physician performed an independent interpretation of the images. At present, there is no additional reimbursement for the use of AI in most clinical scenarios, though this may change in the future.

For readers who want a deeper understanding of how POCUS billing works in general, check out this 3-part POCUS Billing Primer that provides the foundational framework on which AI-enabled POCUS billing is built.¹

Background

Medicine has a long history of adopting emerging technologies that introduce new diagnostic and therapeutic capabilities without an immediately established reimbursement pathway. While these innovations often benefit patients, they also carry costs related to acquisition, implementation, and maintenance. As a result, the development of reimbursement mechanisms is both necessary and consequential.

The American Medical Association, which establishes and defines CPT codes, created Category III CPT codes to address this challenge. The AMA describes Category III codes as “a set of temporary codes that allow data collection for emerging technologies, services, procedures, and service paradigms.² These codes are intended to be used for data collection to substantiate widespread usage or to provide documentation for the Food and Drug Administration (FDA) approval process.”

In practical terms, Category III codes are designed to track the utilization of novel technologies and procedures while evidence is gathered to determine whether they warrant transition to Category I CPT codes (traditional, permanent CPT codes). Category III codes are identified by a four-digit numeric code followed by the letter “T.” When a service is described only by a Category III code, it should still be submitted for billing purposes, although there is typically no guaranteed reimbursement attached. Submission allows utilization tracking and may ultimately support conversion to a Category I code. Importantly, some Category III codes do receive payment when they map to an Ambulatory Payment Classification (APC) under CMS.

Separately, the Centers for Medicare & Medicaid Services (CMS) has established additional mechanisms to support reimbursement for novel technologies, recognizing that appropriate payment structures can both improve patient care and encourage innovation.³ These mechanisms include:

• NTAP (New Technology Add-On Payment) for inpatient services
• Technology Pass-Through (TPT) payments for outpatient hospitals and ambulatory surgery centers
• NT-APC (New Technology Ambulatory Payment Classification) for outpatient clinics³

To qualify, a device or technology must generally be newly FDA-cleared, costly, and demonstrate substantial clinical improvement over existing alternatives. For NTAP specifically, payment is made in addition to the established Diagnosis-Related Group (DRG) payment for an inpatient stay. Companies that believe their technology meets these criteria may apply directly to CMS.

In 2021, Caption Health, an AI-based guidance technology designed to assist with cardiac POCUS image acquisition, received NTAP approval. In 2023, EchoGo, a software platform that detects heart failure with preserved ejection fraction (HFpEF) using cardiac POCUS, was approved for payment through both NTAP and outpatient reimbursement via a Category III CPT code. By late 2025, EchoGo Heart Failure achieved a nationally established outpatient payment rate of approximately $300 under CMS. This marked a notable shift, representing one of the first instances in which AI-augmented POCUS received payment at a level exceeding standard imaging.

More recently, the Health Tech Investment Act, introduced in Congress in April 2025, aims to create a more predictable reimbursement pathway for algorithm-based AI technologies under Medicare. If enacted, the legislation would include algorithm-based AI tools used in conjunction with other procedures, eliminate the need for repeated case-by-case CMS approvals, and establish a guaranteed five-year payment period to allow emerging technologies time to demonstrate value.

Applying This to POCUS

There are now many healthcare products that are FDA-cleared or in development that leverage AI to augment existing POCUS applications or enable entirely new use cases. Despite this growth, only a small subset of these technologies has been approved by CMS for independent reimbursement. At present, the threshold for obtaining a distinct reimbursement pathway remains high.

In most real-world POCUS scenarios, the AI component itself is not separately reimbursed. Instead, reimbursement follows the usual POCUS billing pathway, while AI serves as an adjunct; either improving image acquisition efficiency or generating quantitative measurements that may otherwise be impractical due to time or training constraints.

For this reason, it is essential that the physician independently interprets at least a portion of the POCUS examination. The professional fee is tied to the physician’s cognitive work and interpretation, not to the AI output. When AI is used to assist with quantification or diagnostic suggestions, the report must clearly document physician review and independent interpretation of the images.

AI-generated measurements and diagnostic labels should be treated as decision-support tools rather than definitive interpretations. Reports should reflect the physician’s own assessment, not simply restate AI-generated values. For example, if AI calculates an ejection fraction of 15%, the physician’s interpretation should document a severely reduced ejection fraction based on independent image review. For clinician-performed ultrasound, reimbursement for the professional component (modifier -26), hinges on this professional interpretation, and AI does not replace the cognitive work required for billing the professional component.

At present, there is no pathway for billing AI-only interpretation of POCUS studies without physician review. Regardless of FDA clearance, AI systems function as assistive technologies rather than independent diagnostic providers.

The primary exception remains the use of Caption Guidance, where, under specific ICD-10 diagnoses, the use of Caption’s AI cardiac guidance qualifies for NTAP payment. It is important to note that this payment applies to the facility and does not alter or augment the physician’s professional fee.

If the Health Tech Investment Act becomes law, it may lower barriers for additional POCUS-AI technologies to qualify for independent reimbursement and could accelerate adoption. However, until such changes occur, existing reimbursement principles continue to apply.

Conclusion

AI will undoubtedly play an expanding role in POCUS, improving image acquisition, streamlining workflows, and enabling measurements that were previously impractical at the bedside. From a reimbursement standpoint, however, the fundamentals have not changed. In most clinical scenarios today, AI does not create a separate billable service; rather, it functions as a tool that supports a clinician-performed and clinician-interpreted POCUS exam. Reimbursement remains tied to medical necessity, appropriate documentation, image archiving, and, most importantly, the physician’s independent interpretation of the study. While emerging payment models and legislation may eventually introduce new reimbursement pathways for AI-enabled technologies, clinicians and health systems should approach AI adoption with a clear understanding of current billing rules, compliance requirements, and risk considerations. Used thoughtfully and documented correctly, AI can enhance the value of POCUS without altering its core reimbursement principles: for now.

Bibliography

1. Peachtree POCUS. A Primer on POCUS Billing – Part 1. Peachtree POCUS Blog. Published September 3, 2024. https://peachtreepocus.com/blog/f/a-primer-on-pocus-billing-part-1. Accessed January 12, 2026
2. American Medical Association. CPT® Category III Codes Long Descriptors [PDF]. American Medical Association; Updated December 30, 2025. https://www.ama-assn.org/system/files/cpt-category3-codes-long-descriptors.pdf. Accessed January 12, 2026.
3. Centers for Medicare & Medicaid Services. New Medical Services and New Technologies (IPPS). CMS website. https://www.cms.gov/medicare/payment/prospective-payment-systems/acute-inpatient-pps/new-medical-services-and-new-technologies. Updated 2025. Accessed January 12, 2026.