Researcher Dhafar Dhirar Saeed Al-Jawadi defended her Master’s thesis on Sunday, 6 September 2026, entitled:
“AI-Based Approaches for Genetic Disorder Diagnosis Based on Facial Phenotype”
The thesis focused on developing an automated artificial intelligence-based framework for analyzing facial morphological characteristics, with the aim of supporting the early detection and classification of genetic disorders based on geometric measurements extracted from frontal facial images.
This approach is based on the fact that many genetic syndromes and disorders are associated with distinctive facial features, proportions, and structures, making facial phenotype analysis an important source of supportive diagnostic information.
The study focused on developing structured and curated facial databases. A public database was established containing images of healthy individuals, along with another database containing images of individuals affected by various genetic disorders, covering more than 75 genetic disorders. A balanced database was also created for multi-class classification, with images distributed equally among healthy individuals, individuals with Down syndrome, and individuals with Cornelia de Lange syndrome.
The study examined three main scenarios. In the first scenario, facial geometric measurements were evaluated using an MLP model. The second scenario involved developing a hybrid CNN–MLP model that combines visual features extracted from facial images with geometric facial features.
In the third scenario, the scope of the study was extended from binary classification to multi-class classification, categorizing individuals into three groups: healthy individuals, individuals with Down syndrome, and individuals with Cornelia de Lange syndrome.
The results demonstrated that combining visual information with geometric measurements provides a more comprehensive representation of facial characteristics. At the same time, geometric measurements offer greater interpretability and can be linked to anatomical features used by physicians when assessing genetic disorders.
The findings also showed that relying on facial geometric measurements provides a computationally lighter model compared with fully relying on raw images. This enhances the potential for employing such approaches in the development of computer-assisted medical diagnostic tools and in the early detection of certain genetic disorders.
The discussion committee consisted of:
- Prof. Dr. Raed Rafea Omar — Chair
- Asst. Prof. Dr. Fares Saleh Fathi — Member
- Lect. Dr. Imad Atiya Khalaf — Member
- Asst. Prof. Dr. Mohammed Abdulmuttalib Mohammed — Member and Supervisor
- Asst. Prof. Dr. Ali Adel Shareef — Member and Supervisor
Prof. Dr. Khalid Khalil Mohammed, Dean of the College, attended part of the thesis defense.
We extend our best wishes to the researcher and the members of the discussion committee for continued success and excellence.
