Facial Diagnosis Software Now A Reality!
Brett | Jan. 22, 2019

Rare diseases can be notoriously difficult to diagnose. This can be due to the treating physician not having much awareness or experience with the rare condition or the fact that many genetic syndromes such as Noonan Syndrome (NS) have such a wide variety of presentations. NS has a 50% misdiagnosis rate and here at the Noonan Syndrome Awareness Association (NSAA) we strive to reduce this misdiagnosis rate by improving awareness and knowledge of diagnostic criteria amongst health professionals in Australia and around the world.
NS can be diagnosed using certain clinical criteria and one of the primary criteria are certain facial characteristics pictured below.
However facial features may be mild and can be hard to differentiate between different cultures, genders and ages. Thankfully clinical diagnosis of NS is now a whole lot easier thanks to Face2Gene – an app utilising an image analysing alrogithm called DeepGestalt.
Recent studies show that many facial analysis technologies measured up to the capabilities of expert clinicians in syndrome identification. In fact Face2Gene has an impressive 91% success rate diagnosing NS clinically!
HOW DOES IT WORK?
Deep learning algorithms build syndrome-specific computational-based classifiers (syndrome gestalts or a ‘template’). Proprietary technology then converts a patient photo into de-identified mathematical facial descriptors. The patient’s facial descriptor is compared to syndrome gestalts (or templates) to quantify similarity (or a ‘score’) resulting in a prioritized list of syndromes with similar morphology. The artificial intelligence then suggests likely phenotypic traits and genes to assist in diagnosis and clinical decision making.
HOW DOES IT PROTECT PRIVACY?
Face2Gene is a completely secured and HIPAA compliant platform. Information and photos uploaded by users remain private and accessible only to users (in most cases the patient’s primary health professional). The system automatically extracts de-identified binary information (a “mesh”) from the images uploaded to perform the analysis. Since patient’s personal health information is not shared with any third parties the examining physician may use the same informed consent regularly used for storing digitized photographs, clinical information and family history information.
WHAT DOES THIS MEAN FOR PATIENTS WITH NOONAN SYNDROME?
NS is diagnosed in two ways: Clincial Diagnosis or Genetic/Molecular Diagnosis.
A Clinical Diagnosis is made by using the following criteria:
Face2Gene allows practitioners to score 1A ‘Typical Face Dysmorphology’ with a high degree of confidence.
Genetic testing involves a blood test and having geneticists analyse a patient’s genes for the known mutations responsible for NS.
Genetic testing is still the most accurate and definitive way to diagnose NS. However, before Face2Gene many patients would have to wait many months to see a Paediatrician for clinical diagnosis and then wait again to see a geneticist and undergo genetic testing to confirm NS. The high accuracy of this technology means health professionals can make a Clinical NS diagnosis with a high degree of confidence.
During this time NS treatment protocols and initiatives can be implemented which means earlier intervention and thus better outcomes for individuals with NS. The accessibility of Face2Gene is of particular importance for health professionals working in rural or remote locations whose patients may have an even longer wait to access genetic counselling than their urban peers.
The NSAA began recruiting participants for facial analysis software in April 2017. We are so pleased to see this project become a reality and be helping people with NS in the community. We thank everyone who participated in the research and all of the health professionals who are using Face2Gene and contributing to the algorithm’s continual improvement.
Subscribe to our blog