{"id":8228,"date":"2026-09-11T08:19:50","date_gmt":"2026-09-11T06:19:50","guid":{"rendered":"https:\/\/www.scs.ch\/uncategorized-en-us\/visual-innovation-through-ai-in-ophthalmology\/"},"modified":"2026-09-20T07:38:56","modified_gmt":"2026-09-20T05:38:56","slug":"visual-innovation-through-ai-in-ophthalmology","status":"publish","type":"post","link":"https:\/\/www.scs.ch\/en\/case-studies\/visual-innovation-through-ai-in-ophthalmology\/","title":{"rendered":"Visual Innovation Through AI in Ophthalmology"},"content":{"rendered":"

Initial Situation<\/h2><\/div>

OCT scans allow for the non-invasive recording of the retina’s fine layer structure. OCT is frequently used in clinics to diagnose major eye diseases, such as age-related macular degeneration, glaucoma, or diabetic retinopathy. However, the manual evaluation of three-dimensional image data is time-consuming and requires medical expertise. AI-supported image analysis can help to automatically segment relevant structures and evaluate large quantities of OCT data in a reproducible manner. <\/p>\n<\/div>

\"Diagramm:<\/span><\/div><\/div><\/div>

Project Overview<\/h3><\/div>

Field of Application<\/strong>
Ophthalmology and Research<\/p>\n<\/div>

<\/div><\/div>

Area<\/strong>
Medical & Life Science<\/p>\n<\/div>

<\/div><\/div>

Solution<\/strong>
AI segmentation of medical structures in OCT scans.<\/p>\n<\/div>

<\/div><\/div>

System Environment<\/strong>
On-premise platform, dockerized containers, web interface for anonymized OCT scans<\/p>\n<\/div>

<\/div><\/div>

Key Features<\/strong>
Cross-device OCT segmentation, synthetic tumor images with GANs, reproducible AI development<\/p>\n<\/div>

<\/div><\/div>

Evidence<\/strong>
Results on OCT segmentation published scientifically (PLOS ONE, 2019)<\/p>\n<\/div><\/div><\/div><\/div><\/div>

The Contribution of SCS<\/h2><\/div>

SCS developed a modified U-Net CNN for the automated segmentation of OCT images and set up a reproducible development environment for training, evaluation, and operation. For rare tumor images, SCS additionally investigated methods for synthetic data generation. <\/p>\n<\/div><\/div><\/div>

<\/i><\/div>

U-Net Adapted for OCT<\/h4><\/div>
<\/div>
\n

After evaluating various CNN architectures, SCS opted for a modified U-Net that enables pixel-precise segmentation of complete OCT recordings.<\/p>\n<\/div><\/div><\/div>

<\/i><\/div>

High-Quality Reference Data Established<\/h4><\/div>
<\/div>
\n

For training and comparison, hundreds to thousands of OCT B-scans were annotated with pixel precision. In addition, independent segmentations from different groups of people were used to compare the model’s performance with human evaluations. <\/p>\n<\/div><\/div><\/div>

<\/i><\/div>

Rare Tumor Images Synthetically Generated<\/h4><\/div>
<\/div>
\n

Only limited training data was available for rare tumor images. SCS therefore investigated generative methods and was able to generate more realistic synthetic tumor images with GANs than with the previously tested autoencoder approaches (few-shot learning). <\/p>\n<\/div><\/div><\/div>

<\/i><\/div>

Developed Reproducibly<\/h4><\/div>
<\/div>
\n

Code, datasets, and experiments were consistently versioned. Immutable datasets, a digital lab journal, and versioned final states make it possible to track results for research and publications in a reproducible manner. <\/p>\n<\/div><\/div><\/div>

<\/div><\/div><\/div><\/div>
\"Drei<\/span><\/div><\/div><\/div><\/div><\/div>
<\/div>

Result<\/h2><\/div>

In the scientific evaluation, the automated segmentation of the examined eye structures was within the variability of human evaluations. The model was deployed as a dockerized service on a local platform. Anonymized OCT scans can be evaluated via a web interface, and the areas and volumes of the segmented eye structures can be determined. The approach supports different OCT recording methods and devices. Scientifically published: “Validation of automated artificial intelligence segmentation of optical coherence tomography images”, PLOS ONE, 2019. <\/p>\n<\/div><\/div><\/div>

At a Human Level<\/h3><\/div>

Segmentation of eye structures within the range of manual evaluations.<\/p>\n<\/div><\/div><\/div>

Cross-Device<\/h3><\/div>

The model works for various OCT devices.<\/p>\n<\/div><\/div><\/div>

PLOS ONE 2019<\/h3><\/div>

Scientifically published evaluation of OCT segmentation.<\/p>\n<\/div><\/div><\/div><\/div><\/div>

Services Provided<\/h2><\/div><\/div><\/div>

Algorithms, ML & AI<\/h3><\/div><\/div><\/div><\/div><\/div><\/div>

Computer Vision<\/h3><\/div><\/div><\/div><\/div><\/div><\/div>

Medical & Life Science<\/h3><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
\"Smiling<\/div><\/div>

Your Contact Person<\/h2><\/div>
Beat H\u00f6rmann<\/div><\/div>

Healthtech & Innovation
+41 43 456 16 00<\/a><\/p>\n<\/div>

Speak with the responsible specialist<\/span>Speak with the responsible specialist<\/span><\/div><\/a><\/div>

For questions regarding machine learning, medical image segmentation, and reproducible AI development.<\/p>\n<\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>

Further Case Studies<\/h2><\/div>