Located within the Lariboisière Hospital in Paris since 2020, the EviRed Reading Center was specifically designed to meet the needs of the project. It provides expert, centralized, and standardized interpretation of ophthalmic images collected as part of this prospective cohort study.
With a team of trained graders, the center processes ultra-widefield multimodal images (color fundus photography, OCT, OCT-A), using a standardized terminology to identify, analyze, and localize retinal lesions. This approach ensures traceability of assessments and the creation of a high-quality annotated dataset — a key foundation for the development, training, and validation of the project’s artificial intelligence algorithms.
On a daily basis, the reading center operates with the support and involvement of our partners: Carl Zeiss Meditec, ADCIS, and Evolucare.
The EviRed Reading Center is supported by a multidisciplinary team combining advanced medical and technical expertise.
Image interpretation is conducted by retina-specialist ophthalmologists with recognized experience in multimodal imaging analysis, as well as by orthoptists and clinical research technicians specifically trained in image reading protocols within the framework of clinical trials.
This clinical team works in close collaboration with two data scientists responsible for ensuring the availability and organization of imaging data, with the support of our partners ADCIS and Evolucare.
The center’s operations are overseen by our two medical leads, Dr. Sophie BONNIN and Prof. Pascale MASSIN, who also actively contribute to the training and supervision of the grading team.
The Reading Center is responsible for interpreting various types of imaging data collected during examinations performed in the 3,086 included patients across 14 partner ophthalmology departments in France.
Imaging sessions are conducted every day for patients enrolled in the EviRed cohort. We process images acquired from a variety of devices, including wide-field color fundus photographs, OCT and OCT-Angiography from PlexElite, Cirrus, Spectralis, Optovue and Topcon systems.
To ensure the quality and reliability of our assessments, each image is independently reviewed and annotated by two separate graders. In the event of a discrepancy, the image is submitted to an experienced super-grader for arbitration.
On a monthly basis, an average of 1000 images are reviewed and annotated. Graders work in a blinded manner on anonymized images, ensuring the highest possible level of objectivity in their assessments.
Discover our activity as seen by a grader
I am an orthoptist by training, graduated from Sorbonne University in 2019. I then pursued a Master’s degree in Health, specializing in “Chronic Diseases and Disability – Rehabilitation Research” at the same university. Alongside my Master’s, I gained professional experience in pre-consultation settings within centers and ophthalmology practices.
I was looking for an end-of-studies internship for my second-year Master’s degree, and EviRed had just started. The project was innovative and ambitious, and it felt like the perfect opportunity to take my first steps into research.
My role involves annotating different types of imaging (UWF color fundus photography, OCT, OCTA) based on specific medical criteria. Beyond these tasks, my work also includes training new team members, creating various tutorials, training materials and lexicons, as well as contributing to certain ancillary projects. We also maintain daily communication with our partners, particularly regarding data management and the development of our annotation platforms.
Image quality is a real issue for us; poor quality can lead to misinterpretation or missed lesions. However, these are real-life images necessary for training an AI. Graders annotate using only one modality at a time. For example, with color fundus photography, it has sometimes been difficult in certain cases to distinguish between IRMAs and neovessels. It can be frustrating at times not to have access to OCTA to verify, as we would in routine clinical practice, but that’s part of the process!
Traceability is well ensured: each annotation is timestamped and linked to a user account, and every version is preserved. The quality of annotations is maintained partly through the training of the different graders; a lexicon created by EviRed serves as our guideline. Additionally, a double reading process has been implemented, along with super grading performed by senior ophthalmologists. These quality controls allow us to resolve discrepancies between two graders if needed and to correct any inconsistencies in the annotations.
AI is going to transform, and is already transforming, orthoptics just as it is transforming medicine as a whole. These technologies will play an increasingly important role in our practice. The orthoptist of tomorrow must stay attentive to these advances in order to adapt their practice and evolve it when necessary.
Following a project like EviRed from its beginnings to the very end has been particularly enriching and has greatly changed my usual orthoptist practice. My ability to assess the severity of diabetic retinopathy has improved, as well as my skill in distinguishing between physiological features and pathological signs.
I am much more comfortable with imaging and computer tools and better equipped to find solutions to problems. Working with very different profiles has also helped me develop greater adaptability.
Yes, of course! The goal is to develop an AI tool capable of predicting and identifying patients at risk of progressing to a severe form of diabetic retinopathy within one year, thus enabling personalized follow-up accordingly. Beyond our project, I believe EviRed still has much more to offer. We have created a database with a cohort of over 3,000 diabetic patients followed annually; this data is invaluable.
We are, in a way, the eyes that train the AI, and that’s quite rewarding, isn’t it? The role of a grader itself requires a certain rigor that I appreciate, and we are fortunate to work with cutting-edge multimodal imaging. But beyond that, being part of EviRed means contributing to a concrete scientific advancement that will have a real impact on the clinical practice of tomorrow — at least, we hope! We work in a team coming from very diverse backgrounds and professions, and our exchanges are intellectually stimulating.
Not a specific case in particular, but we’ve made it a regular practice to discuss certain challenging cases or uncertainties in our annotations during team meetings. These conversations are always interesting, and the answers are rarely straightforward. Often, they lead us to refine definitions within our lexicon. The field of research is constantly evolving, and this ongoing progress is what makes the work so stimulating on a daily basis.
Don’t hesitate to take the leap; there’s a place for every profile. You’ll find yourself at the heart of innovation and future medical advances, actively involved in a process that both challenges you and makes a real difference in patient care.
At EviRed reading center, we leverage advanced expertise in OCT-A (Optical Coherence Tomography Angiography) to visualize the retinal microvasculature with unmatched precision. This non invasive imaging technology is central to our mission: improving the detection, monitoring, and understanding of diabetic retinopathy by combining cutting edge innovation with clinical excellence.
| Cookie | Duration | Description |
|---|---|---|
| cookielawinfo-checkbox-analytics | 11 months | This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Analytics". |
| cookielawinfo-checkbox-functional | 11 months | The cookie is set by GDPR cookie consent to record the user consent for the cookies in the category "Functional". |
| cookielawinfo-checkbox-necessary | 11 months | This cookie is set by GDPR Cookie Consent plugin. The cookies is used to store the user consent for the cookies in the category "Necessary". |
| cookielawinfo-checkbox-others | 11 months | This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Other. |
| cookielawinfo-checkbox-performance | 11 months | This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Performance". |
| viewed_cookie_policy | 11 months | The cookie is set by the GDPR Cookie Consent plugin and is used to store whether or not user has consented to the use of cookies. It does not store any personal data. |