OPHTHALMOLOGY
quantusGL: AI-based glaucoma risk assessment
From a retinal image, it uses Artificial Intelligence to analyze fundus images to automatically detect glaucoma risk and provide an assessment within minutes.
DESCRIPTION
quantusGL is a test for glaucoma detection from fundus images.
It is a software medical device (MDSW) designed for use by ophthalmologists to provide a prediction (expressed as a risk percentage) of glaucoma. The device provides medical information to support decision-making in the management of patients at risk of glaucoma.
The generated report is designed to be used as a diagnostic support tool; its results should not replace a complete clinical evaluation performed by a specialist physician.
USE CASES
CU-GL-01: Prioritization of cases for ophthalmological review
When faced with a high volume of fundus images pending review, quantusGL automatically analyzes each image and provides a glaucoma risk prediction, allowing the ophthalmologist to prioritize the case list and review those with the highest risk first.
CU-GL-02: Support for screening in primary care consultations
During a primary care consultation, a fundus image is taken of patients over 40 years old during a routine check-up, and quantusGL analyzes the image, providing a glaucoma risk percentage that the physician uses to assess the need for referral to a specialist, extending screening to a care setting other than ophthalmology.
CLINICAL EVIDENCE
Tool for automatic glaucoma detection based on the automatic analysis of a fundus image (quantusGL)
The aim of the study is to design and evaluate an automatic glaucoma classifier from fundus images. A total of 10,657 images from several public databases were used.
A Deep Learning network was trained for binary classification (glaucoma / normal). Of the 10,657 total images, 7,106 were used for training (with a glaucoma prevalence of 24%), while the remaining 3,551 were reserved to evaluate the model.
The following results were obtained:
| Sensitivity | Specificity | PPV * | NPV * |
|---|---|---|---|
| 84.1% | 95.8% | 52.4% | 99.1% |
* PPV and NPV: Positive Predictive Value and Negative Predictive Value
CLINICAL VALIDATION
Validation of the tool using images from real clinical practice.
To evaluate the correct performance of the algorithm against real-world data, an initial sample of images from real clinical practice was used, obtained from the Institute of Applied Ophthalmobiology (IOBA) at the University of Valladolid, the Valladolid University Clinical Hospital, and the Río Hortega University Hospital (HURH).
The IOBA dataset included 1,158 45-degree fundus images, centered on the macula and focused on the optic disc, corresponding to 1,158 eyes from 616 patients.
During validation, an area under the ROC curve (AUC) of 76% was obtained.
SOLUTION INTEGRATION
- Our products allow integration with client systems through the DICOM protocol and the HL7 FHIR interface
- We ensure data privacy in SaaS installations by establishing IPsec VPNs with our clients.
- All communications outside the client environment are secured with SSL TLS 1.3
LICENSING BY NUMBER OF TESTS
LICENSING BY ANALYSES PERFORMED
CUSTOM OFFERS
