PaiVizion Secures China's First Class III AI Radiotherapy Certification with iCurve E, Paving the Way for Intelligent Oncology Treatment

Since the first AI-assisted diagnostic software incorporating deep learning gained market access in 2020, the National Medical Products Administration (NMPA) has approved nearly 100 Class III medical device certificates for AI products with built-in deep learning capabilities over the subsequent four years, covering a wide range of pathologies in medical imaging.

However, the vast majority of these medical AI products are focused on empowering image diagnosis, with very few digital-intelligence applications available for clinical treatment scenarios.Prior to 2025, only three related applications—joint replacement, pulmonary surgery, and hepatorenal surgery—achieved breakthroughs.

Distribution of Class III Medical Device Certifications for AI in China (as of January 10, 2025)

At the recently held “Tianjin Forum on New Advances in Tumor Radiotherapy 2025” organized by the China Anti-Cancer Association, PVmed and the Department of Radiation Oncology at Tianjin Medical University Cancer Institute and Hospital jointly announced PVmed’s independently developediCurve E Intelligent Radiotherapy Delineation SystemHaving obtained the Class III Medical Device Registration Certificate issued by the National Medical Products Administration, PVmed’s iCurve E Intelligent Radiotherapy Delineation System, based on deep learning, has pioneered the entry into AI-driven radiotherapy, further enriching the applications of AI in the therapeutic field.

According to VCBeat’s verification, this Class III medical device certificate is the fourth in China for AI-assisted therapy, and alsoThe First Class III Medical Device Certificate in the Field of AI Radiotherapy. Currently, this certification can be used for the contouring of thoracic tumors and related organs and tissues prior to the formulation of radiotherapy plans,Marking the Official Entry of China’s Radiotherapy into the New AI Era。

Radiotherapy contouring is the most time-consuming step in the entire radiotherapy planning process and requires the most physician expertise, directly affecting both side effects and final treatment outcomes.

In the clinical workflow, radiation oncologists need to delineate normal organs and adjacent tissues (i.e., Organs at Risk, OARs) using imaging modalities such as CT and MRI prior to treatment, so as to minimize radiation exposure to healthy tissues during radiotherapy while maximizing the dose delivered to the tumor volume.

This is a task that severely tests physicians’ patience; delineating the target volume on a 128-slice CT series requires approximately three days of a physician’s effort. Nowadays, to achieve more precise radiation therapy, CT scans with higher resolution and greater numbers of slices are increasingly used in clinical practice, leading to a sharp increase in the time physicians spend on radiotherapy contouring.

Meanwhile, contouring the radiation therapy target volume also tests the competence and experience of radiation oncologists. Where are the boundaries of a target volume? Which areas within the target should receive high doses, and which should receive low doses? Where are the interfaces between the target volume and normal tissues? These questions require radiation oncologists to maintain a high level of attention throughout the lengthy contouring process.

Faced with the numerous pressures encountered by radiation oncologists during target volume delineation, AI may be one of the best solutions currently available.

Previous retrospective clinical studies have shown that the introduction of AI-assisted tools during radiotherapy delineation can significantly improve delineation accuracy and reduce inter-observer and intra-observer variability. Statistical data from this prospective, multicenter clinical trial demonstrated that “AI + manual” delineation outperformed purely manual delineation.

Meanwhile, AI can also address efficiency issues in radiation therapy contouring. Taking PVmed’s iCurve E Intelligent Radiation Therapy Contouring System, which recently received certification, as an example,iCurve E significantly shortens the delineation of target volumes and organs at risk—a process that typically takes hours or even days in clinical practice—reducing the contouring time from several hours to approximately 10 minutes., significantly shortening the patient waiting period and optimizing treatment workflows.

Overall, integrating AI into radiotherapy contouring not only enhances the quality of contouring, optimizes radiotherapy workflows, and improves diagnostic and treatment consistency, but also liberates physicians from extensive repetitive tasks. This indirectly increases the supply of radiotherapy resources within hospitals, delivering tangible health economic benefits.

Prior to the approval of iCurve E, existing AI tools on the market typically validated their efficacy through retrospective studies. While this approach facilitated market entry for AI technologies, it failed to effectively verify their generalizability across diverse real-world clinical settings in different regions.

Nowadays, PVmed’s iCurve E has obtained approval as a Class III medical device through prospective, controlled, and systematic clinical trials. This achievement not only marks its market access for radiotherapy AI but also systematically validates the clinical value of radiotherapy AI through prospective, controlled clinical trials.

As the world’s first AI-powered radiotherapy product to obtain Class III medical device certification through a prospective, controlled clinical trial, iCurve E’s registration clinical trial was led by Tianjin Medical University Cancer Institute and Hospital, in collaboration with authoritative medical institutions including The Fifth Affiliated Hospital of Sun Yat-sen University, Union Hospital Affiliated to Tongji Medical College of Huazhong University of Science and Technology, Shanxi Hospital of Cancer Hospital Chinese Academy of Medical Sciences, and Guangxi Zhuang Autonomous Region Nanxishan Hospital. At the forum, Yuan Zhiyong, Director of the Department of Radiation Oncology at Tianjin Medical University Cancer Institute and Hospital and Principal Investigator of the clinical trial, delivered a systematic presentation on the clinical study design, implementation process, and core findings of iCurve E.

The entire clinical trial spanned 18 months, and the product officially obtained the Class III medical device registration certificate from the National Medical Products Administration (NMPA) in January 2025. Yuan Zhiyong stated that, as the first multicenter, prospective, controlled, and systematic clinical trial in China to evaluate the accuracy and effectiveness of AI-assisted contouring, this project conducted large-scale validation of AI-assisted radiotherapy contouring technology in real-world clinical settings. The research team carried out a substantial number of AI-assisted physician contouring tasks across multiple centers, established a rigorous multi-round quality control mechanism, and conducted data collection for over six months. Throughout the study, more than 20,000 structure-level data contours were completed, all manually performed by over thirty radiation oncologists, accumulating thousands of hours of work. This study provides high-level evidence-based support for the clinical efficacy and safety of AI radiotherapy contouring products and offers a reference standardized paradigm for the regulatory approval pathway of AI-based therapeutic medical devices in China.

Yuan Zhiyong pointed out that the advantages brought by AI-assisted delineation are “comprehensive.” In terms of accuracy, among the major organs covered in this clinical study, AI-assisted delineation significantly outperformed manual delineation across all structures, with a significant improvement in the average Dice Similarity Coefficient (DSC) metric (p < 0.0001), demonstrating the immense potential of AI in enhancing consistency, standardization, and quality control.

In terms of efficiency, AI-assisted manual delineation reduces the average time required for target volume delineation from 66.4 minutes to 12 minutes, which is only one-fifth of the time required for purely manual delineation. Meanwhile, for every additional 50 CT slices, AI-assisted delineation adds only 2 minutes, whereas manual delineation adds approximately 17.5 minutes. As target volume delineation becomes increasingly refined, the value of AI becomes ever more significant.

Comparison of AI-Assisted Delineation vs. Manual Delineation in Terms of Accuracy and Time Efficiency

Leveraging its first-mover advantage in regulatory registration and approval, bolstered by the authoritative endorsement of prospective, controlled clinical trials, PVmed has significantly accelerated its commercialization efforts.

Currently, PVmed’s business model focuses on two core pathways: first, large tertiary Grade A hospitals and specialized oncology hospitals; second, radiation therapy centers at tertiary and secondary Grade A hospitals in prefecture-level and county-level cities.

Under the current distribution of radiotherapy resources, large tertiary Grade A hospitals and specialized oncology hospitals concentrate the vast majority of radiotherapy services, boasting extensive high-end radiotherapy equipment and seasoned radiation oncologists, making them the primary target for radiotherapy AI companies seeking commercialization.

For large tertiary Grade A general hospitals and specialized oncology hospitals, AI companies such as PVmed deliver value by eliminating inter-observer and intra-observer variability, thereby enhancing quality and efficiency.

Since its inception, PVmed has been deeply committed to the “AI + cancer treatment” sector, from the deployment of its first-generation contouring product for nasopharyngeal carcinoma radiotherapy in 2018, to the launch of the fifth-generation intelligent individualized target volume contouring platform, iCurve Pro, in 2023, and further to obtaining in 2025China's First Class III AI Certification in the Field of Radiation Therapy, the product portfolio and AI technology stack are continuously evolving and upgrading, with sustained improvements in precision, efficacy, generalizability, and safety. Leveraging its intelligent radiotherapy product line covered by three NMPA Class III medical device registration certificates, PVmed’s contouring platform currently supportsOver 200 Organs at Risk and 11 Common Cancer TypesAutomatic Target Volume Delineation, covering multiple complex sites including nasopharyngeal cancer, lung cancer, breast cancer, cervical cancer, colorectal cancer, esophageal cancer, oral cancer, and laryngeal cancer; it supports automated delineation based on CT and MRI, establishing a comprehensive system for whole-body organ-at-risk (OAR) and target volume delineation.

Currently, PVmed is the company in China with the largest number of Class III medical device registrations for radiotherapy contouring software, covering key modules such as image fusion, dose analysis, and automatic target volume delineation. Among the 19 relevant registration certificates published by the National Medical Products Administration (NMPA), PVmed holds three exclusively, firmly securing its leading position in the industry. The recent approval of iCurve E not only marks a significant milestone for AI-driven radiotherapy in China,It has achieved a groundbreaking innovation globally as the first AI-based oncology treatment device to be validated through prospective, controlled clinical trials., providing important samples for exploring the safety, generalizability, trust mechanisms, and clinical value of AI technologies in therapeutic scenarios.

But this represents not only a leap in product performance, but also an evolution in strategic role. PVmed is transforming from a “provider of AI contouring tools” into a “co-builder of smart radiotherapy departments,” leveraging AI as the core driver to assist hospitals in establishing high-quality, efficient, replicable, and evaluable radiotherapy systems. This “second growth curve” no longer relies solely on software licensing or single-function product deployment; instead, it targets comprehensive upgrades in departmental capabilities.

InLarge Grade 3A General Hospitals and Specialized Oncology HospitalsWith a high volume of radiotherapy patients, heavy physician workload, and long waiting times being the norm, iCurve E can reduce target delineation time from hours to minutes, alleviating physician burden and easing radiotherapy bottlenecks. Meanwhile, it enhances standardization and ensures consistency in radiotherapy quality. The expert experience models accumulated by the AI system also provide high-quality references and training pathways for junior physicians, helping hospitals rapidly improve the overall proficiency of their teams.

As forRadiotherapy Centers in Prefecture-Level and County-Level Cities, AI has become a key capability for retaining patients and preventing “upward referral overcrowding.” With PVmed’s AI, even junior physicians can rapidly produce contouring results that meet the standards of top-tier tertiary hospitals, enabling patients to receive higher-quality treatment locally, reducing the burden of referrals, and truly achieving the decentralization and local accessibility of high-quality medical resources.

PVmed is advancing the standardized development of “AI Radiotherapy Departments” across various hospitals, built upon five core modules—“AI Delineation + Image Reconstruction + Intelligent Quality Control + Information Interconnectivity + Online Adaptive Therapy”—and integrated with mainstream international radiotherapy equipment through seamless interface compatibility. Transitioning from mere “usability” to “trusted adoption” and “widespread use,” PVmed serves not only as a pioneer in AI-driven medical products but also as a practitioner and promoter shaping the future of smart radiotherapy in China.

As the “second growth curve” for radiotherapy AI companies such as PVmed, the market corresponding to radiotherapy centers in prefecture-level and county-level cities is burgeoning and offers advantages in scale. Therefore, in the long term, its market potential is even expected to surpass that of large Grade A tertiary hospitals and specialized oncology hospitals, becoming another important growth driver for PVmed.

"After all, decentralizing the radiotherapy capabilities of large tertiary Grade A hospitals and specialized oncology centers through AI solutions will create a win-win situation for all stakeholders.""Standardized Construction of AI Radiotherapy Departments"This approach not only helps non-tier-3 hospitals build new competitive advantages within the existing payment system, but also enables patients who previously had to travel over 100 kilometers for medical care to receive treatment in their hometowns, thereby better supporting long-term battles against cancer.

Since 2017, PVmed has engaged in in-depth collaboration with the Sun Yat-sen University Cancer Center on target volume delineation for nasopharyngeal carcinoma, overcoming challenges in delineating complex anatomical structures, and gradually expanding its applications to multiple radiotherapy scenarios, including head and neck tumors, lung cancer, and cervical cancer.

Following the approval of iCurve E, PVmed has completed AI R&D for radiotherapy treatment planning, target volume delineation, and dynamic dose calculation, securing Class III medical device registration certificates for all these solutions. An intelligent ecosystem centered on precision radiotherapy has thus taken shape.

Furthermore, PVmed’s AI system has been integrated with major international linear accelerator manufacturers, enabling optimization of the online adaptive radiotherapy workflow.

For instance, in its collaboration with Philips on AI-driven radiotherapy solutions for spectral CT, Peking Union Medical College Hospital deployed the multi-parametric functional Spectral CT RT system. By integrating this system with the PVmed AI radiotherapy platform, it can simultaneously analyze over a dozen categories of multi-parametric functional imaging data for individual patients. This enables minute-level automatic delineation of target volumes and organs at risk (OARs) across thousands of high-dimensional thin-slice images, providing an efficient solution for the clinical implementation of spectral CT in precision radiotherapy.

At its benchmark site, Fujian Provincial Cancer Hospital, PVmed has established an intelligent radiotherapy system integrating five components: “AI contouring + CT + MRI + Elekta Harmony Pro adaptive accelerator + quality control platform.” This comprehensive system covers seven wards and 30 physician workstations across the entire center, providing full support for high-difficulty target volume contouring in tumors of the head and neck, pelvis, and other regions. It successfully enabled the province’s first case of AI-guided adaptive radiotherapy for cervical cancer.

Currently, the hospital’s entire radiation oncology ward has achieved 100% coverage of intelligent contouring, with seamless integration between AI-generated contours and physician review, ensuring that every patient receives a standardized, high-quality radiotherapy plan.

Under a multi-dimensional layout, PVmed’s commercial capabilities are even stronger.

Entering a new phase, PVmed aims to further refine its core AI technologies, expand the coverage of AI in radiation therapy to additional cancer types, and accelerate the research and development of minimally invasive oncologic surgery, thereby enhancing its independently developed AI platform for minimally invasive surgical procedures. For instance, Tianjin Medical University Cancer Institute and Hospital pioneered the integration of AI-assisted target volume delineation into its breast cancer radiotherapy workflow. By combining this with the “Tianzhong Expert Model,” the hospital has significantly improved the efficiency of treatment planning, providing a practical model for the intelligent transformation of specialized oncology hospitals.

In terms of commercialization, PVmed is accelerating its global market expansion. Currently, the company has initiated the implementation of AI-driven oncology treatments in Southeast Asia, the United States, and Europe, leveraging its international expansion to establish diversified revenue streams for its AI solutions.

Overall, PVmed’s achievements and future plans have set a benchmark for the development of medical technology enterprises in China, providing a replicable, evaluable, and scalable pathway model for the construction of “AI-powered smart radiotherapy departments.”

From filling the gap in global prospective clinical trials of AI in radiotherapy to paving the way for radiation therapy into the intelligent era, PVmed is advancing steadily onto the global stage, contributing a “Chinese solution” to promoting the intelligent transformation of cancer treatment.