Characterization and Recognition of Genetic Diseases by Photography

There are around 8,000 rare diseases and new ones are described every month in the scientific literature. They affect a limited number of patients. Nearly 80% of these diseases have a genetic origin and 30 to 40% of them are associated with dysmorphia. The latter can be suspected by evaluating the morphological characteristics of the patient. This medical skill, called dysmorphology, which allows a diagnosis to be made by evaluating the morphological characteristics of a patient, is based on experience. Diagnosis is often easy for relatively common diseases, but more difficult for rarer pathologies affecting few patients and often described in a single ethnicity and age of life.

The study aims to create a dataset specific to the application of methods from artificial intelligence. Extending the methodologies described to profile and extremity photographs will allow better recognition and description of dysmorphia.

This will allow to make diagnostic suggestions by comparison with the database. The Data Science team has already explored the notion of phenotypic similarity of patients.

Jean Feydy is a mathematician expert in image analysis and will ensure the scientific robustness of the study methods.

This project will conclude with the establishment of a diagnostic aid tool, integrating research results for doctors with a particular interest in developmental anomalies and intellectual disability.