Seed project: Using Artificial Intelligence to Detect Acanthamoeba Keratitis in the eye - the AIDAK study Applicants

Project information

Title
Using Artificial Intelligence to Detect Acanthamoeba Keratitis in the eye - the AIDAK study Applicants
Applicants
Linnaeus University: Patrick Bergman, Jenny Roth, Antonio Macedo, Alisa Lincke and Welf Löwe
External collaboration: Neil Lagali, Linköpings University
Project period
November 2022 – June 2023
Core research areas
Ophthalmology, eHealth/health informatics, computer science
Financier
Linnaeus University Centre for Data Intensive Sciences and Applications (DISA)
Proposal
AIDAK såddansökan DISA.pdf
Final report
Slutrapport DISA sådd AIDAK.pdf
Presentation
Presentation AIDAK_disa_ehealth.pdf
Research groups
E-health – Improved Data to and from Patients
Data Intensive Software Technologies and Applications (DISTA)

Photo: Per Fagerholm, Linköpings universitet

More about the project

The overall objective of the research is to develop a platform to assess whether a person is likely to have AK based on the confocal microscopic images. The objective of this seed project is to create a network of researchers that jointly takes initial steps towards this objective.

Acanthamoeba are ubiquitous organisms found in the air, soil, drinking water, tap water, swimming pools, and in both saltwater and freshwater environments. They can cause corneal infections that lead to Acanthamoeba keratitis (AK) (Figure 1). AK is a serious medical condition accompanied by intense eye pain, sensitivity to light, and severely impaired vision. If not treated within a few weeks after infection, the parasite causes permanent corneal damage, necessitating subsequent transplantation. If allowed to further progress, all eye structures can eventually become infected, resulting in a total loss of the eye (1). Contact lens wear-related infection is the most frequent etiology of AK, accounting for up to 90% of all cases. Given their popularity particularly in Scandinavia, with almost 15% of adults in Sweden wearing contact lenses, AK, once considered rare, is becoming more common.

Rapid identification and diagnosis of AK is paramount for preserving vision. Currently all diagnosis of AK using IVCM is done by an experienced human observer and there are many borderline cases. Here computer vision (CV) using Artificial Intelligence (AI) can provide a solid platform to quickly, in a cost-effective manner, and with higher reliability detect and rule out cases with AK.

The overall objective of the research is to develop a platform to assess whether a person is likely to have AK based on the confocal microscopic images. The objective of this seed project is to create a network of researchers that jointly takes initial steps towards this objective. More specifically:

1. Based on an already existing image database at Linköping University Hospital we will create a dataset with labeled data. The image database consists of around 70000 images of infected and suspected cases of infection, and several hundred thousand images of healthy eyes and other diseases.
2. Conduct a pilot study to train, adapt, and evaluate standard CV deep learning networks to detect AK based on images from confirmed AK cases.
3. Write a manuscript and submit to a suitable journal
4. Write a grant proposal, firstly to FORSS, but other suitable funders will be identified.

Expected Results:
The main outcome of this seed project is expected to be a project group that will work together with the main aim of securing funding to develop an AI-based platform that can be used with clinically obtained IVCM images to provide a decision support tool to assist treating physicians in assessing the probability of AK infection, to guide subsequent treatment. Secondary results include one publication in a peer reviewed journal of relevance and a presentation at an international conference.

DISA's seed projects

Learn more about the seed project concept and DISA's other seed projects.