Automatic Phenotyping of Patients on 2D Photography

Automatic Phenotyping of Patients on 2D Photography

The field of artificial intelligence is booming in medicine and in the field of diagnosis. The data can be varied: x-rays, pathology sections, or photographs.

It is considered that 30 to 40% of the 7000 rare diseases described to date cause craniofacial dysmorphia. Their detection sometimes requires the trained eye of a geneticist, because certain phenotypic traits are subtle. These diagnostic difficulties and the fact that certain diseases are extremely uncommon lead to considerable diagnostic delays

The patient inclusion criteria are:

• Patients followed in medical genetics,
• Patients undergoing maxillofacial surgery, or craniofacial surgery as part of the management of a pathology, of genetic origin or not, associated with dysmorphism of the head and neck,
• Patients for whom frontal and profile facial photographs are taken as part of their treatment.

The inclusion criteria for control subjects are:

• Patients followed in maxillofacial surgery, for a disease other than a rare disease associated with dysmorphia in the head or neck: acute pathology (wound) or chronic (gynecomastia).
• Patients for whom frontal and profile facial photographs are taken as part of their treatment.

The criteria for non-inclusion of patients are:

• Patients who have undergone facial or skull surgery before the first photo was taken.
• Person subject to a judicial safeguard measure.
• People objecting to the reuse of their health data.

The criteria for non-inclusion of control subjects are:

• Pathologies affecting facial symmetry (dental cellulitis, displaced fractures).
• Patient followed for dysmorphic syndrome or in whom dysmorphic syndrome has been suspected.
• Person subject to a judicial safeguard measure.
• People objecting to the reuse of their health data.