12 December 2016

Diagnosis of genetic diseases by photo

Dmitry Ryder, XX2 century, based on the materials of MIT Technology Review: Diagnosing Disease with a Snapshot Thousands of doctors and researchers dealing with genetic diseases around the world are already using a computer program for preliminary diagnosis 


FDNA, the parent company Face2Gene, was created six years ago after the Israeli co-founders sold the former company for the production of facial recognition software Face.com companies Facebook. This program is able to distinguish between specific faces after "training" on several images of a specific person. Program Face2Gene, on the contrary, defines a pattern common to a group of people with the same syndrome. The establishment of this common denominator allows the program to create a composite characteristically averaged image correlated with the disease.

When CEO Dekel Gelbman was hired in 2010, he met with several medical practitioners and quickly realized that facial recognition could help reduce the number of undiagnosed diseases.

According to the assessment Face2Gene of more than 7000 known genetic syndromes, up to half are associated with a special type of face that can be recognized and used to establish a diagnosis. Down syndrome, for example, is one of the most common, and therefore it is easier to diagnose. But rarer diseases can prove challenging.

 Так выглядит профиль загруженного фото пациента с оценкой вероятности одного из множества заболеваний, представленных в базе программы

"Imagine that you are a geneticist and you have your own archive of images in your head corresponding to what you have seen and what you have been taught," Gelbman explained. — You can't know everything. Your ability to conjure up an image or try to compare models in your mind is limited. How do you simplify it?"

Face2Gene uses data collected by geneticists for free. Although the technology is free for information providers, Gelbman provides for payment by pharmaceutical companies for access to assistance in drug discovery and development. The more people enter data into Face2Gene, the more the system remembers facial features associated with any syndromes. According to Gelbman, 60 percent of clinical geneticists and genetic consultants worldwide use this technology.

The mobile application automatically photographs the patient, uploads this photo to the server, and analyzes facial features within a few seconds to compile a list of syndromes corresponding to identified similarities. Each syndrome is accompanied by information from London medical databases, in which a collection of dysmorphic images is stored and updated. 

Users can superimpose their patient's face on the face of a prototype with the disease and click on a heat map showing the areas of the face that most closely match the representative image of a particular syndrome. They can also note the signs present in their patient, thereby recalculating the list of potential syndromes.

In the future, FDNA will continue to work on determining facial features related to various diseases, in particular, autism. "We want to know if we can find a face for the fragile X chromosome," Gelbman said, referring to the most common hereditary cause of mental retardation in boys. "Let's see what we can find."

The application is designed to identify diseases mainly in children. Its use is free of charge, but it is assumed that it will be used exclusively by medical specialists and researchers. When registering, it requests a medical specialty, qualification and the name of a medical or scientific institution.

To diagnose a patient's genetic diseases, it is enough to upload one or more of his front photos using web applications or applications for Android or iOS. Each processed request can be saved in the program database, but there is no need to worry about confidentiality: it is not the person itself that is saved, but his deindividualized mathematical model (formalized phenotype).

Based on the results of comparing the uploaded photo with the types of faces characteristic of certain genetic diseases, the program gives a list of probable diseases of the patient and shows on the scale the degree of probability of the disease — from low to high.

Perhaps the application can be useful to you if you are a pediatrician or are studying congenital diseases.

Doctor Karen Gripp, Head of the Department of Medical Genetics at the Alfred Dupont Pediatric Hospital (Nemours Alfred I. DuPont Hospital for Children) in Wilmington, Delaware, noticed that her patient, a little girl, had a high forehead, thick eyebrows and a long filtrum — a groove between her nose and upper lip. She suffered from epileptic seizures and had coarse curly hair, although it would seem that this is hardly worth noting because of her African-American ethnicity. The cleft palate plus the arachnoid cyst on her spinal cord had already been operated on by the time Flu examined her.

The symptoms do not line up exactly in line with the diseases that the Flu was considering.

Many genetic diseases have a "face" — a distinctive combination of features that give the key to a potential diagnosis. But the detection of the disease through observation of the patient's characteristics, a practice known as dysmorphology, is one of the biggest challenges for geneticists. The most experienced dysmorphologists, as a rule, are elderly doctors who have seen a lot, which is quite understandable: the more patients you see in your life, the more signs you observe. But even the most experienced practitioners are not able to notice every disease.

As part of her research, Flu asked permission to take a picture of the patient and upload the photo to Face2Gene, a facial recognition program that can help diagnose rare diseases. Face2Gene compares images of the patient's face with "sketches" of the disease and gives a number of potential diagnoses, from the most to the least likely.

For Flu, the technology turned out to be a revelation. The Hyde — Cheney syndrome was one of the main assumptions Face2Gene in the case of this child. But exomic analysis of protein-coding genes revealed a mutation associated with lateral meningocele syndrome, a disease with which Flu was more than familiar. She wrote a scientific article about him.

Harelip is not a sign of lateral meningocele syndrome, so Flu initially did not consider it as a diagnosis. But his resemblance to the Hyde—Cheney syndrome helped her to pay attention to him. Since few patients have this diagnosis, the girl's future is unclear. She will survive, but her condition will be closely monitored due to possible neurological problems, as well as scoliosis, which are strongly associated with the disease.

"I kept linking the disease to the harelip, but it was completely a false trail," said Gripp, who believes that the harelip is not related to the genetic disease of the girl. — You only recognize what you know. Face2Gene made me go through the list and think, "Hmm, did I take everything into account?""

Portal "Eternal youth" http://vechnayamolodost.ru  12.12.2016

Found a typo? Select it and press ctrl + enter Print version