AI: A Boon, Not a Bane, for Australian Health Imaging Sector
Artificial intelligence (AI) has undoubtedly captured the global spotlight, sparking intense discussions among investors and analysts alike. This surge in interest has led to significant market volatility, particularly within the software sector, as the potential disruptive power of AI is meticulously weighed. The emergence of sophisticated AI coding tools has fuelled speculation that software development could become faster, cheaper, and more accessible, prompting a re-evaluation of traditional Software-as-a-Service (SaaS) business models.
However, within Australia’s burgeoning health imaging sector, the sentiment surrounding AI appears to be shifting from apprehension to opportunity. Leading ASX-listed health imaging firms are increasingly viewing AI not as a threat to their established operations, but as a powerful catalyst for growth and efficiency.
Pro Medicus: Caught in the Tech Crossfire
Pro Medicus (ASX:PME), a long-standing success story on both the S&P/ASX 200 Index and the S&P/ASX 200 Health Care Index, has experienced considerable share price volatility recently. Despite a robust first-half financial performance, the company’s stock has seen a significant decline year-to-date. This downturn is attributed, in part, to broader market concerns about tech valuations and the potential impact of AI.
Pro Medicus CEO and co-founder Sam Hupert, however, has publicly addressed these anxieties, arguing that the company has been inadvertently caught in the wave of massive capital expenditure directed towards AI infrastructure development. He asserts that Pro Medicus operates a capital-light, software-only model, suggesting that the company stands to benefit from the very infrastructure investments that are currently driving market apprehension.
Hupert also strongly refutes the notion that AI can easily replicate complex, enterprise-grade software like their flagship Visage 7 platform. He emphasises that Visage 7 is built on proprietary technology, incorporating over three decades of specialised domain knowledge. This, he argues, is not a product that can be easily reproduced, with or without AI, and that Pro Medicus has deliberately avoided creating a simple roadmap for competitors. Furthermore, he highlights that their offering extends beyond mere software, encompassing integrated systems and methodologies that enable the delivery of highly sophisticated solutions to leading global medical institutions.
The Enduring Importance of Radiologists and Enterprise Imaging
Concerns that AI might render radiologists obsolete are also being challenged. Hupert views this as an oversimplification of current AI capabilities within diagnostic imaging. He points to manpower surveys suggesting that efficiency gains from AI are likely to be offset by an increasing workload driven by new imaging modalities and larger datasets. This means AI will likely serve to augment, rather than replace, the expertise of radiologists, who are already in high demand.
Iain Wilkie, a healthcare analyst at Morgans who covers Pro Medicus, echoes this sentiment. He highlights the structural constraints within radiology, where the growth in CT and MRI volumes has consistently outpaced the supply of radiologists for years, leading to bottlenecks, reporting backlogs, and burnout.
Wilkie champions AI as a potent tool for enhancing workflow efficiency but stresses that it does not replace the core function of radiology. Crucially, he adds, AI is dependent on the robust enterprise-grade imaging backbone that companies like Pro Medicus provide. This underlying infrastructure, encompassing secure routing, comprehensive data management, advanced visualisation capabilities, audit trails, and compliance layers, is precisely where Pro Medicus excels. As Wilkie aptly puts it, they provide “the rails the system runs on.”
Navigating a Rigorous Regulatory Landscape
The healthcare IT sector, particularly radiology information systems (RIS) and enterprise imaging workflows, operates within a highly regulated environment. Wilkie points out that stringent regulations, such as HIPAA in the US (mandating strict privacy, encryption, audit trails, and role-based access) and GDPR in Europe (governing personal health data), are paramount. The incoming EU AI Act further classifies AI-enabled radiology tools as high-risk, necessitating human oversight, conformity assessments, and post-market surveillance.
In Australia, the Therapeutic Goods Administration (TGA) regulates any technology used for serious disease diagnosis. Similarly, the US Food and Drug Administration (FDA) classifies RIS and associated workflow software as Class II medical devices, subject to rigorous quality system regulations. Globally, hospitals must adhere to standards set by medical bodies like the American College of Radiology for accreditation.
This complex web of regulations means that clinical infrastructure cannot be rapidly or cheaply replaced. Procurement cycles for such systems can span years, as seen with Pro Medicus and Mach7, where contracts often take a significant amount of time to implement. Healthcare institutions are inherently risk-averse, and any new AI-driven challenger must overcome substantial hurdles related to security, explainability, liability, and reimbursement – challenges that are orders of magnitude greater than those faced by consumer-focused AI applications.
Singular Health: Strengthening with AI
Singular Health Group (ASX:SHG) sees AI not as a disruption but as an enhancement to its existing technologies. CEO Denning Chong confirms that AI is already integral to the company’s 3DICOM technology and will become increasingly important. However, he clarifies that Singular Health is focused on building a comprehensive infrastructure around imaging workflows, rather than developing a singular AI algorithm.
Chong explains that 3DICOM serves as a workflow and interoperability platform, facilitating patient access, education, and clinical use. He believes that the advancement of AI tools will actually bolster Singular Health’s position. As more FDA-regulated AI models become available, the demand for a compliant platform capable of seamlessly integrating these tools into clinical workflows will grow. Singular Health’s marketplace is designed to accommodate multiple imaging AI applications within a single, secure infrastructure.
The company is actively addressing a persistent challenge in medical imaging: the fragmented nature of image access, visualisation, and sharing across disparate healthcare systems. Traditional systems often store medical images in DICOM format, managed by Picture Archiving and Communication Systems (PACS) that operate in isolation. This fragmentation hinders image sharing and contributes to costly duplicate scans.
Singular Health’s 3DICOM technology acts as a crucial bridge between these systems. It integrates a Medical File Transfer Protocol, a Volumetric Rendering Platform, and AI in the Cloud, enabling secure transfer, viewing, and analysis of imaging data. Chong identifies the secure interoperability and regulated deployment of imaging data as the more significant challenge than algorithm development itself. He asserts that Singular Health provides the secure infrastructure – the “rails” – upon which imaging data moves, and as AI adoption expands, the value of this foundational infrastructure is poised to increase.
EMVision: AI as an Accelerator
EMVision Medical Devices (ASX:EMV), a developer of portable brain scanners, also views AI as a significant accelerator of its development process, rather than a competitive threat. CEO Scott Kirkland highlights that AI is instrumental in speeding up various aspects of development, from advanced simulations to the testing and debugging of code.
EMVision is currently conducting pivotal validation trials for its first commercial device, the emu, designed for rapid stroke detection and classification at the patient’s bedside. The company is also developing the First Responder, a lighter, backpack-sized version intended for use in ambulances and at emergency scenes to diagnose strokes and, in the future, traumatic brain injuries.
Kirkland notes that many common AI solutions in healthcare are trained on vast datasets of existing medical images (CT, X-ray, ultrasound, MRI). While this offers a faster path to market for well-resourced AI companies, it can also make it difficult to establish a long-term competitive advantage.
EMVision, in contrast, is generating unique, device-specific clinical datasets through its brain scanners, which utilise electromagnetic signals rather than conventional CT or MRI techniques. These proprietary datasets, comprising unique electromagnetic scattering measurements of the brain, do not exist in the public domain. This approach creates a multi-layered defensibility for EMVision: a physical moat through proprietary hardware, a data moat via exclusive datasets, and an algorithm moat by training its own AI models on this unique data. This entire ecosystem is further protected by a robust intellectual property portfolio.
Kirkland concludes that AI significantly enhances EMVision’s development cycles. The complex signals processed by their hardware are exceptionally well-interpreted by AI, leading to highly reliable diagnostic results at the point of care. The company’s edge, he states, lies in its ownership of the hardware, the data, and the AI models.
