Visual innovation through AI in ophthalmology

AI Innovation Measurement technology Medical technology

Optical coherence tomography (OCT) is a gentle laser microscope and is used by ophthalmologists to examine the retina for abnormalities. Trained AI models support doctors in recognising pathologies.

  • Initial situation

    OCT scans can be used to record the fine layered structure of the retina. However, the evaluation of 3D scans for tumours, for example, is time-consuming and has not been tried and tested to date. Artificial intelligence (AI) can support doctors in this process.

  • SCS solution

    SCS trained a modified U-Net-CNN model for the detection of eye layers and tumours. The accuracy achieved is equivalent to that of experienced ophthalmologists.
    The model runs dockerised as a service on an on-premise platform.

  • Added value

    The AI solutions automate and accelerate the detection of eye layers and tumours in 3D OCT scans. The model works for various OCT devices. This means that the AI can be used by many ophthalmologists in research.

Detection of defects in the eye

Due to its intrinsic 3D resolution and non-invasive ease of use, OCT is widely used in clinics to diagnose major eye diseases, including age-related macular degeneration, glaucoma, diabetic retinopathy and vitreo-retinal traction.

Artificial intelligence (AI) can help doctors to recognise these pathologies in OCT scans more efficiently. The AI-supported findings are analysed by doctors and reduce the time a doctor spends in front of the computer. And they potentially reduce the diagnostic error rate.

Labelling data

Labelling the eye layers requires a great deal of expertise – and a lot of time. Several experts labelled hundreds to thousands of OCT-B scans pixel by pixel. SCS stored these in versioned form.

OCT images are obtained from the eye, labelled by experts, and the U-Net is trained using these pixel maps.

Decision in favour of the architecture of the modified U-Net

As a basis, research was carried out into the CNN architectures used for semantic segmentation of medical images. The U-Net has proven itself and still achieves state-of-the-art results, i.e. the resulting error corresponds to the irreducible error. In other words, the U-Net is indistinguishable in its accuracy from ophthalmology experts.

Results of segmentation without tumour images

The project comprises several studies and analyses. The one described below is limited to an illustrative evaluation without tumour images. The “human factor” plays a major role in the results. Even humans do not produce a 100% (see line 1 in the table below) consistent ground truth (GT). It therefore makes sense to compare the variations in segmentation between different people. Hamming Distance was used as a measure for comparing performance. The U-Net performs better than the heterogeneous group of labellers consisting of laypersons, opticians and professional ophthalmologists.

ComparisonVariation
Within GT (1 expert, 3 times repeated labelling)1.3-1.7%
GT and CNN1.4-1.7%
10 people (amateurs, opticians, ophthalmologists)1.7-3.5%

Reproducibility

The software code is managed in a Git repository. Data sets are versioned and stored “immutably” on a network drive. Experiments are kept in a digital lab journal and final versions (for publication[i] or production) are also versioned. Reproducibility is extremely important in this project, as the results are used for publications.

Rollout and operation

The backend runs in an on-premise infrastructure in dockerised containers and will be available to the medical community in the future. Users upload queries including their anonymised OCT scans via a user-friendly web interface. The model then determines the volume and areas of the eye layers and makes the results available to the user via a web interface.

Few-Shot Learning

A particular challenge is the small amount of tumour data within the images. In a further study, various techniques for the artificial generation of ground truth data were trialled. Approaches using autoencoders delivered blurred results. Only the use of Generative Adversarial Networks (GANs) allowed reliable artificial generation of tumour images.

Three OCT images of an eye tumour, generated with a GAN. The area surrounding the tumour shows a high light intensity. You can also see a shadow under the tumour in the OCT images.

Summary

Provided that sufficient, i.e. hundreds to thousands of high-quality and pixel-by-pixel labelled OCT scan images are available, the U-Net recognises medically relevant structures with an accuracy comparable to human ophthalmology experts.


[i]Validation of automated artificial intelligence segmentation of optical coherence tomography images
Maloca PM, Aaron Y Lee, Emanuel R de Carvalho, Mali Okada, Katrin Fasler, Irene Leung, Beat Hörmann, Pascal Kaiser, Susanne Suter, Pascal W Hasler, Javier Zarranz-Ventura, Catherine Egan, Tjebo FC Heeren, Konstantinos Balaskas, Adnan Tufail, Hendrik PN Scholl Scholl (2019) Validation of automated artificial intelligence segmentation of optical coherence tomography images. PLOS ONE 14(8): e0220063

Related projects

Visual inspection of railway wagons

Is the block brake worn? How thick is the pantograph contact strip? Is an incorrectly fitted screw coupling hanging down? In future, SBB's "Visual ... More

Reliable AI for the measuring wheel set

SBB uses so-called measuring wheel sets to measure the dynamic contact forces between rail and wheel. Sensors on the wheels measure the forces. It is ... More

Smart Meter Toolkit

According to the Electricity Supply Ordinance, the grid operator must enable the end customer to receive real-time measurement data from the smart ... More

Medical spectrometer with electro-optical circuit board

An electro-optical spectrometer has been miniaturised so that it can be used in the confined spaces of an intensive care unit. The device monitors ... More

IOBnet – Exchange platform for ophthalmologists

The IOBnet platform offers ophthalmologists the opportunity to exchange anonymised cases with experts at any time. By creating the platform, SCS is ... More
Show all projects
Beat Hörmann Healthtech & Innovation How can I help you?