MEDICAL

A use case - macular holes

A macular hole is an opening in the retina (the layer at the back of the eye that is sensitive to light) that develops at the fovea (the part of the eye that is responsible for sharp vision) and causes a small dark spot in the central vision. Often preventing those with the condition from recognising very small objects, particularly from reading ordinary print and possibly being fit to drive.

For people over the age of sixty idiopathic macular holes are an important cause of visual loss. In this population 0.5% will have macular holes and about two thirds of these are women. While most people only develop a macular hole in one eye there is a 10 to 15% chance of developing a macular hole in the other eye as well.


Problems with measuring macular holes

Current macular hole measures are typically made using a single two-dimensional (2D) slice of a horizontal optical coherence tomography (OCT) image and measured by a human grader using callipers. This is known to be prone to high intra and inter-observer error and vulnerable to further error from off- centre scan location.

When studied in three dimensions (3D) macular holes are found to have significant asymmetry in all dimensions. Human measurements in 2D of these asymmetric macular holes have considerable inter-observer variability and fail to represent key parameters of the macular hole accurately. Intogral is used by retinal eye surgeons and could be used by ophthalmologists. It gives reliable macular hole measurements and a better understanding of what the macular hole looks like in 3D, so that they can quantify the hole better.

Intogral's AI platform

Intogral utilises cutting-edge research in Image Analysis, Deep Learning, to learn from the expertise of experienced clinicians. These expert clinicians, with decades of experience help our software to learn how to spot diseases in high-resolution medical images, augmenting the ability of clinicians to diagnose and treat their patients.

Intogral's Platform is accessed through a User Interface (UI) which keeps in the style of existing OCT viewers that clinicians will be familiar with and introduces a brand-new 3D view along with 3D measurements. Retinal eye surgeons can use these enhanced measurements to improve the management planning and outcome prediction for the treatment of macular holes.


Intogral's 3D image annotation platform

Our 3D deep learning models produce rapid and high-quality 3D image data, using Intogral's cloud-based Image Annotation Platform.

A use case - macular edema (diabetic retinopathy)

Macular edema is the build-up of fluid in the macula, an area in the center of the retina. The retina is the light-sensitive tissue at the back of the eye and the macula is the part of the retina responsible for sharp, straight-ahead vision. Fluid buildup causes the macula to swell and thicken, which distorts vision.

Macular edema occurs when there is abnormal leakage and accumulation of fluid in the macula from damaged blood vessels in the nearby retina. A common cause of macular edema is diabetic retinopathy, a disease that can happen to people with diabetes. Macular edema can also occur after eye surgery, in association with age-related macular degeneration, or as a consequence of inflammatory diseases that affect the eye. Any disease that damages blood vessels in the retina can cause macular edema.


Intogral's AI platform

Intogral utilises cutting-edge research in Image Analysis, Deep Learning, to learn how to spot diseases in high-resolution medical images, augmenting the ability of clinicians to diagnose and treat their patients.

Publications

  • A. V. Nasrulloh, C. G. Willcocks, P. T. G. Jackson, C. Geenen, M. S. Habib, D. H. W. Steel, and B. Obara, Multi-scale Segmentation and Surface Fitting for Measuring 3D Macular Holes,IEEE Transactions on Medical Imaging, vol. PP, no. 99, PP. 1–1, 2017, ISSN: 0278-0062. DOI: 10.1109/TMI.2017.2767908.
  • Chris G. Willcocks, P. T. Jackson, C. J. Nelson, A. Nasrulloh, and B. Obara, Interactive GPU Active Contours for Segmenting Inhomogeneous Objects, Journal of Real-time Image Processing, Dec. 2017, ISSN: 1861-8200, DOI: 10.1007/s11554-017-0740-1.