RetinaLyze’ new AI algorithm detects signs of glaucoma using retinal images

A team of researchers have developed an algorithm, which automatically detects abnormalities and changes linked to the eye disease glaucoma. The new algorithm is now offered as a part of the software, RetinaLyze, which provides instant eye screening results for the 3 most frequent diseases causing blindness glaucoma, macular degeneration and diabetic retinopathy.

Posted on
June 18, 2018
Company news

Blindness can be avoided

Glaucoma is the
leading cause of irreversible blindness in the world, but blindness from
glaucoma can often be prevented with early treatment. The eye disease damages
your eye’s optic nerve and usually happens when the fluid inside the eye
(aqueous humour) has difficulty circulating. 
Mainly, the difficulty for its exit increases the pressure in your eye,
damaging the optic nerve. Assessing the risk of getting glaucoma is a critical
first step towards reducing the likelihood of a patient suffering from its blinding

“In 2013, the number
of people with glaucoma worldwide was estimated to be 64,3 million, increasing
to 76 million in 2020 and to 111,8 million in 2040."[1]

Time-consuming and
expensive to screen accurately for glaucoma today

To perform
this assessment, doctors consider multiple parameters/risk factors such as
symptoms, genetics, family history, other diseases and medication. However, the
only sure way to screen for glaucoma is with a complete eye exam which includes
several examinations of the eye. These examinations utilize hardware such as
visual perimetry and OCTs, which are considered time-consuming and expensive
for screening purposes.

A common
and easily available method of screening for glaucoma is measuring the internal
eye pressure, which in many cases causes the blinding damage to the optic nerve
head. While the method is economical, it’s not very accurate as use of
tonometry on its own results in a high number of false positives and because it
is possible to have glaucoma without it affecting the eye pressure.

Until now,
assessing glaucoma using a fundus/retinal image of the optic nerve head has
been a time-consuming and complicated task reserved for ophthalmologists due to
the many parameters involved and the necessary background knowledge.

An objective, fast and
economical alternative

A team of
Spanish researchers decided to work on an objective, fast and economical method
for glaucoma screening. Fundus cameras are relatively accessible devices, both
in terms us and procurement, and therefore an algorithm was developed for
instant analysis of a simple color photograph of the retina of the eye.

algorithm, named RetinaLyze Glaucoma, compares the color of the optic nerve to
the color of the veins and arteries of the retina, and deduces its degree of
perfusion. This process is carried out in 24 optic nerve sectors, since
perfusion defects caused by glaucoma affect some areas more than others. The
joint analysis of the information obtained in these 24 areas, in more than 2500
cases of normal and abnormal subjects, makes
it possible to obtain a mathematical function that defines the normal eye by
looking at the distribution of the Blood Globin Protein (GDF) and detecting
abnormal cases, such as glaucoma.

In layman’s
terms, the algorithm uses machine learning (AI) to assess the damage to a vital
part of the eye (the optic nerve head) to deduce if a patient should be
referred to a manual assessment for glaucoma.

The RetinaLyze Glaucoma AI algorithm assesses the Optic Nerve Head of a fundus photo to detect signs of Glaucoma

to the main researcher of the project and clinical software developer, Prof.
Manuel González de la Rosa M.D. Ph.D., it is “

As reliable as human
examination and other screening methods

Manuel González de la Rosa, who has led the project, has highlighted the
"objectivity" of this detection method, which only requires the
patient to take photographs of the retina of the eye in a hospital, clinic and
even an optometrist store setting.

algorithm has been clinically validated in studies involving over 2500 cases
and has found to be as reliable as human examinations of fundus images. The
algorithm has also been compared with alternative methods such as assessment
using OCT or Visual perimetry, where it achieved similar and better

In a study
involving 7 different fundus cameras and over 1.800 normal and glaucomatous
cases, the algorithm reached a specificity of 95,5 % and a sensitivity of 86,5
% in detecting glaucoma at a stage that can be confirmed by other methods.

Available today in

algorithm has been made commercially available through the platform,
RetinaLyze, in May 2018 after over a year of clinical and commercial
validations and test performed by a group of doctors, optometrists and

already offers screening algorithms for signs of Diabetic Retinopathy and AMD
(Age-related Macular Degeneration) and tele-medicine services to eye-care
specialists such as ophthalmologists and optometrists. The platform
automatically presents instant results of suspicious lesions in three colors
that indicate whether there is basis for a referral to an eye specialist
(green, yellow, and red).

Commercial Officer and co-founder of RetinaLyze System A/S (Ltd.), Ganesh Ram,
welcomes the new cooperation and adds: “

About RetinaLyze System A/S (Ltd.)

RetinaLyze System A/S is a medical
technology company established in 2013, which is represented in 35 countries
all over the world. The company delivers groundbreaking decision support
tools to eye-screening professionals to enable efficient, accessible and fast
eye-screenings. All our algorithms and services are CE-marked, clinically
validated and patented.

Ganesh Ram
Chief Commercial Officer
Tel. +45 25 94 44 08

Tel. +45 42 70 12 03



Ganesh Ram

CCO and Co-founder of @RetinaLyze.
Productivity and UX-geek.
Passionate about making an impact.