The Neural Network Was Taught To Recognize 216 Rare Hereditary Diseases By Photography - Alternative View
The Neural Network Was Taught To Recognize 216 Rare Hereditary Diseases By Photography - Alternative View
Video: The Neural Network Was Taught To Recognize 216 Rare Hereditary Diseases By Photography - Alternative View
Video: Pediatric and Rare Disease - Mort Cowan Diana Farmer Rosa Bacchetta and Pawash Priyank 2023, February
Anonim

Researchers have developed an artificial intelligence system that can diagnose 216 rare hereditary diseases from photography with high accuracy. As reported in Nature Medicine, she was trained to recognize a genetic disorder (choose from the 10 most likely options) with 91 percent accuracy. Scientists have also made it easier to use the system in practice: they have created a mobile application for doctors that allows you to identify a genetic disorder from a photograph of a patient.

Diagnosing a hereditary disorder is often difficult. There are several thousand diseases associated with genetic disorders, most of which are extremely rare. Many doctors during their practice may simply not be faced with such diseases, so a reference computer system that would help to recognize rare hereditary diseases would facilitate diagnosis. Researchers have already created similar systems based on face recognition, but they have been able to identify no more than 15 genetic disorders so far, while the accuracy of recognizing several diseases did not exceed 76 percent. In addition, such systems sometimes could not distinguish a sick person from a healthy one. At the same time, the training sample often did not exceed 200 photos, which is too small for deep learning.

Therefore, American, German and Israeli scientists and employees of the FDNA company, under the leadership of Yaron Gurovich from Tel Aviv University, developed the DeepGestalt facial recognition system, which made it possible to diagnose several hundred diseases. Using convolutional neural networks, the system divides the face into separate 100 × 100 pixel fragments and predicts the likelihood of each disease for a particular fragment. Then all the information is summarized and the system determines the probable disorder for the person as a whole.

DeepGestalt splits the face in photographs into separate fragments and evaluates how they correspond to each of the diseases in the model. Based on the aggregate of fragments, the system makes a ranked list of possible diseases. Y. Gurovich et al. / Nature Medicine, 2019
DeepGestalt splits the face in photographs into separate fragments and evaluates how they correspond to each of the diseases in the model. Based on the aggregate of fragments, the system makes a ranked list of possible diseases. Y. Gurovich et al. / Nature Medicine, 2019

DeepGestalt splits the face in photographs into separate fragments and evaluates how they correspond to each of the diseases in the model. Based on the aggregate of fragments, the system makes a ranked list of possible diseases. Y. Gurovich et al. / Nature Medicine, 2019.

The researchers trained the system to distinguish a specific inherited disease from a number of others. For training, they used 614 photographs of people suffering from Cornelia de Lange syndrome, a rare hereditary disease that manifests itself, among other things, in the form of mental retardation and congenital malformations of internal organs. The authors used over a thousand other images as negative controls. DeepGestalt differentiated Cornelia de Lange syndrome from other diseases with 97 percent accuracy (p = 0.01). The authors of other studies achieved 87 percent accuracy, while the experts made the correct diagnosis, on average, 75 percent of the cases. In another experiment, scientists used 766 photographs of patients with Angelman syndrome ("Petrushka syndrome"), which, among other things, is characterized by chaotic movements,frequent laughter or smiles. The system recognized the disease with an accuracy of 92 percent (p = 0.05); in the previous study, the accuracy was 71 percent.

The researchers also taught the system to recognize different types of the same hereditary disease using the example of Noonan syndrome. There are several types of this disorder, each of which is caused by mutations in a particular gene and each has subtle differences in facial features (such as sparse eyebrows). Using a sample of 81 photographs, the authors of the article taught the DeepGestalt system to distinguish between five types of this disease with an accuracy of 64 percent (p <1 × 10-5).

In total, for training the system, scientists used a total of 17,106 photographs representing 216 hereditary diseases. The researchers tested the effectiveness of DeepGestalt on 502 photographs of patients who have already been diagnosed, and on another sample of 329 photographs of patients with a known diagnosis from the London Medical Database. The system determined the patient's disease from the 10 most probable variants with an accuracy of 91 percent (p <1 × 10-6).

The researchers also made it easier for DeepGestalt to be put into practice - they created a platform for diagnosing hereditary diseases by phenotype, as well as a mobile application for doctors, Face2Gene, with which a doctor can diagnose his patient.

Promotional video:

Last year, researchers created a system to automatically recognize plants from their images in herbariums. The convolutional neural network has learned to identify plants with 90 percent accuracy.

Ekaterina Rusakova

Popular by topic