Evaluation of a computer-based facial dysmorphology analysis algorithm (Face2Gene) using standardized textbook photos

Area of Science:

• Medical genetics
• Ophthalmology
• Bioinformatics

Background:

• Genetic syndromes frequently involve ocular manifestations.
• Ophthalmologists face challenges in recognizing dysmorphic features during genetic syndrome assessments.
• Digital image analysis software may assist in identifying these features.

Purpose of the Study:

• To evaluate the sensitivity and specificity of Face2Gene (F2G) software.
• To assess F2G's accuracy in analyzing patient photos from genetics textbooks for dysmorphic feature identification.

Main Methods:

• Analysis of 353 facial photos from two genetics textbooks using F2G.
• Inclusion of variables such as photo type, patient demographics, and disease characteristics.
• Assessment of F2G accuracy, sensitivity, and specificity based on established disease categories.

Main Results:

• F2G correctly identified the exact textbook diagnosis in 42.5% of cases and included it in the top three differentials for 54.1%.
• High sensitivity was observed for craniosynostosis syndromes (80.0%) and syndromes with major facial defects (77.8%).
• F2G demonstrated high specificity (>83%) across all analyzed disease categories.

Conclusions:

• Face2Gene (F2G) is a valuable tool for pediatric ophthalmologists.
• It assists in formulating differential diagnoses for children presenting with dysmorphic facial features.
• The software aids in the evaluation of genetic syndromes with potential ocular involvement.