The European Patent Office (EPO) has granted European Patent No. 4432220 for the invention Annotation of Slice Images in Volumetric Medical Images, developed by a research team from VNUHCM鈥揢niversity of Science (HCMUS). The patent was officially published in the European Patent Bulletin on 8 July 2026. The achievement recognises the research capability, technological innovation, and international intellectual property protection efforts of Vietnamese scientists in the field of artificial intelligence for medical applications.
The inventors are Dr L锚 Kh谩nh Duy, Lecturer at the Faculty of Information Technology; Assoc. Prof. Tr岷 Minh Tri岷縯, Vice President; Nguy峄卬 H峄 Th膬ng Long, alumnus of the Talented Bachelor Programme (Cohort 2018) and researcher at the Software Technology Laboratory (currently a doctoral researcher at Dublin City University, Ireland); and Ph岷 Minh Kh么i, alumnus of the Talented Bachelor Programme (Cohort 2018) (currently a doctoral researcher at Dublin City University, Ireland).
In medical diagnosis, CT and MRI scans are commonly stored as hundreds of individual image slices. To create training datasets for artificial intelligence, specialists must annotate anatomical structures or tumours accurately on every slice. Such work is highly time-consuming, repetitive, and susceptible to inconsistencies between adjacent slices. The process therefore requires considerable effort while maintaining a very high level of precision.
The invention Annotation of Slice Images in Volumetric Medical Images introduces a method for assisting the automatic annotation of 3D medical images by exploiting the spatial relationships between consecutive image slices. Once a user creates or edits an annotation on one slice, a machine learning model predicts and propagates the annotation to neighbouring slices, eliminating the need to repeat the same task manually for every individual image.
A key innovation lies in the ability of the model to analyse not only the content of each image independently but also the visual characteristics of the target structure together with the relative positions of adjacent slices in order to predict the shape of the object throughout the complete 3D volume. As a result, annotations remain more continuous and accurate than approaches that process each slice separately. The system also assigns a confidence score to every annotation, enabling users to identify quickly which slices require further review or refinement. When an annotation is corrected, the system updates and propagates the revised information across other slices, thereby maintaining consistency throughout the dataset. Such capabilities substantially reduce the time and effort required for annotating 3D medical images while producing training data with greater consistency and quality for the development of artificial intelligence systems supporting medical image analysis and diagnosis.
The process of securing a European patent took approximately three years and involved several rounds of examination with specialists from the European Patent Office. According to Dr L锚 Kh谩nh Duy, the greatest challenge was demonstrating novelty and inventive step in accordance with European patentability requirements. The research team developed a new interactive approach between users and medical images, enabling faster and more efficient medical data annotation than existing solutions. To satisfy patent examination standards, however, the interaction mechanism had to be formulated as a clearly defined technical solution that fulfilled the requirements of the European patent system. Previous experience gained by Dr L锚 Kh谩nh Duy while working with human鈥揷omputer interaction research groups at technology companies in Europe also proved valuable in addressing technical issues and supporting the patent prosecution process.
A notable aspect of the achievement is that the invention was conceived and filed independently by Vietnamese researchers at a time when, in 2023, international patent filing support programmes remained limited. The research team explained that pursuing patent protection at an early stage reflected a determination to safeguard an idea with strong practical potential, together with continuous encouragement and support from Assoc. Prof. Tr岷 Minh Tri岷縯 throughout the development process.
Beyond intellectual property protection, the patent application process also enabled the research team to gain valuable experience in intellectual property law, international patent examination procedures, and the preparation of technical patent documentation. According to the researchers, such experience will benefit scientists seeking to transform research outcomes into intellectual assets eligible for international patent protection.
Dr L锚 Kh谩nh Duy also emphasised the importance of interdisciplinary research in achieving patentable outcomes. When a core technological solution is developed within a clearly defined application context, the technical value, practical potential, and overall strength of the invention can be demonstrated more convincingly during patent examination.
Following this achievement, the research team continues to pursue projects in virtual reality and augmented reality, with potential applications in physical rehabilitation training and other simulated environments. Further developments are expected to lead to additional patent applications in the future.
The granting of a European patent for Annotation of Slice Images in Volumetric Medical Images represents a significant milestone for research, innovation, and intellectual property activities at VNUHCM鈥揢niversity of Science. The achievement demonstrates that university research can extend beyond scholarly publication to become internationally protected technological solutions, bringing Vietnamese knowledge and research capability closer to the global scientific and technological community.
Leave a Reply