AI Used to Detect Rare Autoimmune Eye Disease
An artificial intelligence (AI) model has been used to help diagnose thyroid eye disease (TED) – a rare autoimmune condition where tissue including muscle around the eye becomes inflamed. TED can cause a condition called compressive optic neuropathy which occurs when the optic nerve is damaged by compression from the inflammation.
RELATED:
- Novel Targets and Studies in Immunology and Autoimmunity: Antigen-Specific Immunotherapy and Therapeutic Chaperones
- Immunology: from Cancer to COVID
The model analysed CT scans of the eyes of patients from Massachusetts Eye and Ear, a treatment and research centre in Boston, USA. Researchers at the centre presented their findings at the Association for Research in Vision and Ophthalmology’s 2023 annual meeting (ARVO 2023).
Paul Zhou, a research fellow at Mass Eye and Ear, was lead author of the study which claimed that AI techniques could be used to automate screening for TED and help patients receive care faster.
The study used hundreds of images from CT scans of patients’ eyes. 20 eyes were used as control and had no disease; 60 eyes had TED but no compressive optical neuropathy; and 64 had severe TED and signs of compressive optical neuropathy.
Overall the results of the investigation reported an accuracy of 94.27%. The errors that the model made were two control images misdiagnosed as having TED, and six TED-presenting images characterised as being healthy.
“The raw dataset included a total of 885 cross-section 2D images from the seventy-two CT scans in coronal view, which was resampled and adapted to focus on the eye as the region of interest. ” said the study’s abstract.
The AI Model
The Mass Eye and Ear team used an AI model called VGG16, a convolutional neural network that’s used for image recognition. VGG16 is pretrained on the ImageNet dataset which consists of over 14 million images
RELATED:
VGG16 was created by two members of the University of Oxford’s Visual Geometry Group at the Department of Engineering Sciences: Karen Simonyan and Andrew Zisserman. The model was unveiled at the 2014 ImageNet challenge in a paper titled Very Deep Convolutional Networks For Large-Scale Image Recognition. The model was successful in the challenge, coming first in the localisation category and second in the classification category.
Thyroid Eye Disease
Thyroid eye disease (TED) is an autoimmune disease which causes immune cells to attack the orbital fat and connective tissue of the eye. Inflammation of the tissue around the eye can push the eyeballs forward and, in some cases, cause blindness by putting pressure on the nerve behind the eyeball. The disease most commonly occurs in patients with Graves’ disease, the most common cause of hyperthyroidism.
See the upcoming events in Oxford Global’s Immuno series here.
Get your weekly dose of industry news?here?and keep up to date with the latest?‘Industry Spotlight’ posts.?For other Immuno content, please visit the?Immuno Content Portal.
Related Resources
