Your images, your formats
- DICOM, PNG, JPEG and TIFF accepted directly
- Multi-slice DICOM studies, greyscale or RGB, with slice navigation
- Format, integrity and OCT-modality validated on upload
AI-based solution
for ophthalmologists
Automated analysis of OCT and fundus images that returns segmented biomarkers, a classified diagnosis and a ready structured report — in under a minute, in your PACS or in our cloud.
What it looks like in use
A clinician uploads a scan, gets the markup and a finished report that only need checking and signing — then it goes straight into the record.
- The deviceAn OCT scanner or fundus camera in your own room
Our cloud, your server, or a gateway inside your PACS- The patient recordThe report returns to the clinician and the hospital system
platform.visiomed.ai
Every study in one list: type, clinician, status, whether biomarkers were found. A new one is added by dropping a file — DICOM or an image: .dcm, .png, .jpg, .jpeg, .bmp, .tiff.
Open the platformBiomarker segmentation
Every lesion is outlined on its own: switch a biomarker off to see the structure underneath. Beside it, how many lesions were found and how large they really are.
What each modality actually delivers
OCT and fundus run different models, so their conditions and biomarkers differ. Figures for both modalities are validated by leading ophthalmology clinics.
ТУ 58.29.32.000-001-79959792-2025 · ISO 13485
- 14
- Diagnostic classes
- Norm
- early AMD
- middle AMD
- late AMD
- DME
- ERM
- RVO
- RAO
- CSC
- VMT
- VMA
- FTMH
- LMH
- PVD
- 16
- Segmented biomarkers
- IRF
- SRF
- PED
- SHRM
- HIRP
- CDr
- RDr
- DL
- ORA
- TUB
- VPHM
- HIRL
- FTMH
- LMH
- HL
- OPC
- 58s
- Typical analysis timeAveraged across studies in mixed PNG, JPEG, TIFF and DICOM, covering segmentation, classification and report generation.
- 730+
- Studies per dayThroughput for a single deployment. Cloud installations process several studies in parallel and scale with the compute you give them.
Input formats
- DICOM
- PNG
- JPEG
- TIFF
Report formats
- DOCX
It has to fit the clinic you already run
The three questions every medical IT department asks first — what it reads, where it runs, and who vouches for it.
Deployed where your data lives
- SaaS on platform.visiomed.ai — nothing to install
- On-premise inside your own network, images never leave it
- Parallel processing that scales with available compute
Fits your PACS, not the reverse
- DICOM gateway acting as SCP and SCU over C-STORE
- Auto-routing of OPT-modality studies, no doctor action needed
- Results returned as DICOM Secondary Capture overlays and a Structured Report, read in your usual viewer
Built to medical-device standards
- Developed under ГОСТ IEC 62304, ISO 14971 and ISO 13485
- ISO 13485 conformity confirmed by primary production inspection
- JWT authentication, role-based access, TLS and a full audit trail
Who is this for?
Private clinics
Equipment manufacturers
Government clinics
Universities and research centers
Educational ophthalmology centers
Diagnostics and screening
Biomarker segmentation, classification and a finished report for an OCT or fundus image. The clinician gets layers and prose that only need checking and signing.
- Buyers
- Clinics and hospitals · optical and medical chains · healthcare marketplaces · screening programmes
- Model
- Per-clinic subscription · pay per scan on marketplaces · contracts for screening campaigns
Predictive models
Predicting outcome from a series of scans: response to anti-VEGF therapy and the choice between injections and surgery. The value is in the trajectory, not in one image.
- Buyers
- Pharmaceutical companies · clinical trial operators · large clinics and chains
- Model
- Project contracts and a licence for the research environment
Clinical Copilot
A second opinion and an assistant: it searches a patient's studies, cross-references the findings and drafts the conclusion.
- Buyers
- Clinics and chains · medical platforms · clinicians directly
- Model
- Per-seat subscription · API for platforms
Partners
What ophthalmologists say about us
The VisioMed.AI solution is not only a different approach to diagnosing eye diseases but also a significant reduction in the doctor's routine duties, which together allows us to focus on what's most important – our patients and their health.
A trained neural network can be a good assistant for quick diagnosis. But most importantly, it is useful in the face of an impending staffing shortage. Let me explain: tomographs are available in both private and public clinics. But only a small number of ophthalmologists specializing in retinal diseases can work with and understand the images.
As part of the international research group 'Retinalbiomarkers', we see enormous potential for AI in ophthalmology. The use of AI in OCT analysis not only improves diagnosis but also automates the detection of pathologies, increases the accuracy and speed of image interpretation, leading to more effective patient treatment.
Together with VisioMed.AI, we are developing artificial intelligence for ophthalmosurgeons to improve diagnostic accuracy and better treatment strategy selection for macular holes and other retinal diseases. This solution minimizes subjective errors, automates image analysis, and accelerates clinical decision-making.
During the development of VisioMed.AI solutions, doctors manually train artificial intelligence algorithms, which allows AI to adopt expert knowledge and adapt to various clinical scenarios. This approach provides an opportunity to get a second opinion on the presence of a disease, reduces the likelihood of diagnostic errors, and increases the objectivity of OCT image analysis.
The VisioMed.AI solution will become an indispensable tool for novice ophthalmologists, helping them to better understand the structure and pathologies of OCT images. It is not just a diagnostic tool but also a powerful educational platform that contributes to the development of young specialists' professional skills.
Publications
Our researchers hold posts at leading universities worldwide and publish in peer-reviewed journals — the science behind the product is open to inspection.
Segment Anything in Optical Coherence Tomography: SAM 2 for Volumetric Segmentation of Retinal Biomarkers
OCTDL: Optical coherence tomography dataset for image-based deep learning methods
Artificial intelligence for detection of retinal toxicity in chloroquine and hydroxychloroquine therapy using multifocal electroretinogram waveforms
Team
Researchers and engineers in computer vision, working alongside practising ophthalmologists — every model is trained on real clinical judgement, not just labelled data.
Mikhail Kulyabin
Researcher
Ph.D., Kaggle Expert
Postdoctoral Researcher, AI expert
Over 7 years in DL / CV
Ilya Kovrizhnyh
Executive Director
Head of KAM
Project Director in the B2B ecosystem
14+ years in corporate banking:
risk participation in international and Russian transactions
Our achievements
These placements are from public competitions with independent judging, not internal metrics.
See it run on your own scans
Tell us what you image and we will set up a pilot on your data — in our cloud, or entirely inside your network.
- Pilot runs on your own studies, not on a canned demo set
- On-premise option where no image ever leaves your network
- PACS integration handled by our engineers