The structural imbalance of high-quality medical resources represents the most challenging supply-side issue in the current healthcare system. Top-tier hospitals are overcrowded, while primary care institutions face a 'brain drain', and the decade-long training cycle for senior diagnosticians cannot keep pace with the surging demand driven by aging populations and chronic diseases. This fundamental reality underpins the sustained policy support and capital investment flowing into medical AI. Medical imaging is a fast-moving frontier for AI application, characterized by large-scale data, high-frequency diagnostics, standardized workflows, and a structural shortage of relevant physicians. Connecting disparate imaging data, expert knowledge, model training capabilities, and clinical application scenarios has become the key to competition in the industry's next phase.
On July 4th, DIAGENS-B (ASX: 02526) held the "Medical AI Ecosystem Innovation Seminar and iMedLoop Global Medical Imaging Data Platform Launch" in Beijing, officially introducing its latest product: the iMedLoop global medical imaging data platform. The event brought together over a hundred representatives from government, industry, academia, research, and healthcare institutions, including the Chinese Academy of Sciences, the Chinese Academy of Engineering, the China Association of Health Industry and Enterprise Management, the China Academy of Information and Communications Technology, Zhejiang Provincial Cyberspace Administration, Zhejiang Cancer Hospital, Ruijin Hospital affiliated with Shanghai Jiao Tong University School of Medicine, Hangzhou Data Group, and Legend Holdings.
The launch focused on medical imaging data, foundational medical imaging models, specialized disease and discipline AI applications, clinical collaboration, and industry ecosystem development, marking a significant step for DIAGENS in advancing the platformization, scaling, and ecosystem development of medical imaging AI.
iMedLoop: Ushering in an Industrialized, Full-Process Era for AI in Medicine
The iMedLoop platform released by DIAGENS is not a single AI tool but an infrastructure designed for the entire lifecycle of medical imaging AI development and application. It covers key stages including data governance, expert annotation, model training, model service, and clinical application. Its goal is to enhance the development efficiency of medical imaging AI applications and propel specialized disease models from project-based development towards standardization, scalability, and continuous iteration.
If the Ford assembly line transformed automobiles from luxuries for the few into accessible transportation for the masses, then the industrialized production line for medical imaging AI ultimately aims to solve the problem of making high-quality medical services accessible to every household. From an industry logic perspective, past medical imaging AI development often followed a "one-disease-one-model" approach. Different diseases, departments, and imaging modalities typically required re-organizing data, re-annotation, re-training, and re-validation, leading to long development cycles, high costs, and limited cross-scenario replication capability. As clinical demands continue to expand, this single-point model approach struggles to meet the application needs of multi-disease, multi-department, and multi-tier medical institutions.
The core problem iMedLoop seeks to solve is upgrading medical imaging AI from "single model development" to "platform-based model production." Through unified processes for data governance, expert collaboration, model training, model management, and application deployment, the platform can support the incubation and transformation of more specialized disease models, improving the efficiency of moving medical AI applications from research and validation to clinical implementation.
At the data governance level, medical institutions have accumulated vast amounts of imaging data over time. However, whether this data can be used for AI training depends on multiple steps: authorization, desensitization, cleaning, structuring, quality assessment, and security management. iMedLoop incorporates data governance as a foundational capability, helping to transform scattered, dormant imaging data into manageable, annotatable, trainable, and traceable data resources.
In terms of expert collaboration, medical AI training heavily relies on high-quality annotation. Image annotation is not simple data processing but the structured expression of a doctor's clinical experience and diagnostic judgment. Through standardized annotation and quality control processes, the platform can codify expert experience into reusable medical knowledge, further enhancing model training quality and clinical credibility.
At the launch event, DIAGENS also introduced a new-generation intelligent annotation tool, iMedStudio, equipped with four core capabilities: multi-modal fusion, human-machine collaboration, precise segmentation, and intelligent arbitration. It aims to address four common pain points in global annotation tools: messy formats, inefficient manual annotation, inconsistent accuracy, and difficulties in multi-person collaboration quality control.
Regarding model training, DIAGENS has built the iMedImage® foundational medical imaging model, claimed to be the world's first trillion-parameter cross-modal medical imaging large model. It supports 19 imaging modalities, covering over 90% of clinical scenarios, breaking through the limitations of traditional single-modal technology and serving as an underlying engine for developing specialized disease models. According to Song Ning, Chairman of the Board and CEO of DIAGENS, based on this foundational model, the data annotation requirements for training specialized disease models have been reduced to 1/200th of previous needs, the development cycle shortened to 1/12th, and both development costs and computing power expenditure reduced to 1/10th. Leveraging this model, DIAGENS has participated in 6 national and provincial-level major projects and collaborated with 87 top-tier hospitals over the past 12 months to train 145 vertical models.
Based on the medical imaging foundational model, hospitals, research teams, and industry partners can perform adaptive development on top of existing model capabilities, thereby lowering the barrier to developing specialized models and shortening development cycles.
At the model service level, medical AI application is not a one-time delivery. As clinical scenarios evolve, data continuously accumulates, and regulatory requirements increase, models need ongoing evaluation, invocation, management, upgrading, and iteration. The significance of iMedLoop's platform capabilities lies in enabling models to provide continuous service within a unified system, rather than being confined to a single software version or one-off project delivery.
For application scenarios, iMedLoop can support various use cases including imaging-assisted diagnosis, pathological analysis, interventional medicine, report quality control, research analysis, specialized disease screening, county-level medical alliances, and primary care-assisted diagnosis and treatment. AI plays a supportive role, helping doctors complete clinical work by improving efficiency, consistency, and traceability. Final diagnosis responsibility remains with the doctor; the value of AI lies in enhancing the accessibility and service efficiency of medical expertise.
Policy Tailwinds and the Commercial Flywheel: From "Selling Products" to "Selling Capability"
"Artificial Intelligence + Healthcare" has become a crucial direction for medical digitalization. Relevant policies aim to establish a batch of high-quality health industry datasets and trusted data spaces by 2027, forming a series of clinical vertical large models and intelligent agent applications for specialized diseases. By 2030, the goal is for secondary and above hospitals to widely adopt applications like intelligent medical imaging-assisted diagnosis and clinical decision support. This signals that the medical AI industry is moving from pilot demonstrations to scaled application, further elevating the importance of platform-type infrastructure.
DIAGENS's launch of iMedLoop aligns with this trend. By connecting data, experts, models, and scenarios via a platform, it aims to foster stronger ecosystem synergy during the large-scale implementation of medical imaging AI. For hospitals, the platform helps lower the barrier to developing specialized AI applications. For research institutions, it improves the efficiency of converting medical imaging data. For industry partners, it provides foundational capabilities for model development, application validation, and commercial transformation.
Founded in 2016, DIAGENS has long focused on developing medical imaging products and services, forming a portfolio covering medical imaging software, medical equipment, reagents, and consumables. Its core product, AI AutoVision®, is a chromosome karyotype-assisted diagnostic software for intelligent karyotype analysis, intended for prenatal diagnosis of birth defects and assisted reproductive diagnosis. It was the first domestic chromosomal AI product to enter the National Medical Products Administration's green channel for innovative medical devices, boasting a clinical recognition accuracy of 99.86%, 100% sensitivity and specificity for numerical anomaly detection, and compressing traditional 30-day diagnosis times to 4-7 days. In 2024, it captured a 30.6% market share, breaking the international monopoly of German companies Zeiss and Leica and securing the top position in China's chromosome karyotype analysis field. AI AutoVision® can automatically segment, count, arrange chromosomes, and perform case-level anomaly detection on karyotype digital images, helping doctors improve karyotype analysis efficiency. This field has high professional barriers, requiring not only algorithmic capability but also medical knowledge, clinical validation, and medical device registration experience. DIAGENS's accumulation in this area provides a business foundation for its expansion into medical imaging AI platformization.
Beyond its core product, DIAGENS also offers AutoVision® chromosome analysis software, MetaSight® automated cell microscopy image scanning systems, KayoFlow® automated cell harvesters, KayoFlow® integrated slide preparation and staining machines, and various reagents and consumables. These products cover scenarios like chromosome analysis, cell sample processing, microscopic image scanning, and assisted reproduction, forming a relatively complete reproductive health and cytogenetics product system.
From a platform business perspective, DIAGENS has already launched the iMed MaaS® platform, providing end-to-end services from data upload and processing to model training and publishing, supporting zero-code construction of medical imaging model training workflows. The iMedLoop platform is now officially open. According to launch data, over 3,000 professional annotators have joined the platform, with 28.95 million high-quality data entries and over 100 medical AI models deployed. Multiple data element suppliers, AI medical technology companies, and other ecosystem partners are actively participating in platform co-construction.
The launch of iMedLoop can be seen as a further upgrade based on the company's iMedImage® foundational model and iMed MaaS® platform, pushing the company from a product-oriented enterprise towards a platform-oriented one. Commercial data also sends a clear signal: in the first nine months of 2025, DIAGENS's technology licensing revenue reached 57.367 million yuan, accounting for over half of total revenue; annual revenue and gross profit showed significant growth, with gross margin improving simultaneously. Technology licensing and platform services are growing into new core growth drivers, rather than being mere adjuncts to equipment sales. This qualitative change in the revenue structure signifies that the company's value anchor is shifting from "selling products" to "selling capability," moving towards sustainable services.
The competitive barriers in medical imaging AI are also shifting concurrently. The window for first-mover advantage based solely on algorithmic precision is narrowing. The true competitive barriers now stem from a composite cycle involving data assets, expert networks, clinical validation, regulatory compliance, model iteration, and ecosystem collaboration. The platform capabilities DIAGENS is building through iMedLoop are expected to create a "snowball" effect across these areas. Future business models may further expand from single product sales to technology licensing, model services, joint development, specialized model incubation, medical device transformation, and ecosystem partnerships.
According to its prospectus, as of September 30, 2025, the company's products covered over 400 medical institutions across 31 provinces in China, with a 40% penetration rate among the top 100 domestic hospitals. Prestigious institutions like Peking Union Medical College Hospital and Zhongshan Hospital affiliated with Fudan University are among its clients. Currently, DIAGENS has collaborated with multiple hospitals to incubate cutting-edge specialized imaging models covering multiple human organs and disease areas, with related products already deployed in hundreds of medical institutions. This indicates the company is not merely showcasing technology but gradually forming a closed loop from model incubation and clinical validation to commercial transformation. As platform capabilities strengthen, more specialized disease models are expected to be developed, validated, and promoted within the unified system.
For the capital market, the significance of the new iMedLoop product launch lies in DIAGENS's business narrative evolving from a "medical imaging product company" to a "medical imaging AI platform company." The company possesses both a product foundation in niche scenarios like chromosome karyotype analysis and extension capabilities through its medical imaging foundational model, MaaS platform, and technology licensing business. If the platform ecosystem continues to expand, the company's future growth potential could be further unlocked.
The Goal is Enhancement, Not Replacement: Democratizing High-Quality Medical Expertise
The goal of medical AI has never been to replace doctors but to enable the codification, reuse, and dissemination of high-quality expertise. The higher the quality of the AI, the deeper the required involvement of experts and clinical validation. Its ultimate proposition concerns equity: Can a mother in a remote mountainous area and a mother in a first-tier city have an equal chance of receiving an accurate interpretation of the same image? Just as the Ford assembly line brought automobiles to millions of households, medical imaging AI aims to solve the accessibility problem of high-quality medical expertise—empowering primary care doctors with assistive references, allowing expert knowledge to serve more cases, and enabling patients to avoid long journeys for a single imaging diagnosis.
Starting with chromosome karyotype analysis, DIAGENS now uses iMedLoop as a lever to propel medical imaging AI from being a "blessing for the few" towards becoming a "standard tool for the many." The rising technology licensing revenue and deployment in hundreds of hospitals reflect the encoding, replication, and distribution of expert knowledge, reaching corners previously difficult to access. iMedLoop connects dormant data, expert knowledge, and large models, building sustainable infrastructure for producing medical intelligence, transforming sealed medical treasures into health value that can be researched, validated, and shared. As AI becomes an amplifier of medical compassion, we are witnessing not just the transformation of a company, but an era's response to health equity—enabling quality diagnosis to break through walls, traverse mountains and seas, and reach everyone in need.
Comments