Dysmorphic Syndromes Classification and Recognition with Computer Assisted System

Merve Erkınay, Ziya Telatar, Osman Eroğul, and Yusuf Tunca


Dysmorphic Syndromes, Image Processing Techniques , Artificial Neural Network, Leave One Out Method


Genetic diseases, depending on the type of disease is caused by disturbances in the body with different morphologial diseases. These types of diseases are called dysmorphic syndrome. Some of the faces of patients with dysmorphic syndromes show morphological differences compared with healthy people's faces. Based on these differences in purpose of this study is to realize computer-based classification of dysmorphic syndromes. In this study, using image processing techniques, feautures were obtained from facial photographs of patients with Hurler's syndrome, Fragile X syndrome, Prader Willi Syndrome and healthy human face photographs. Feature vectors for each facial image were created. The obtained feature vectors are used to train an artificial neural network and from face images healthy, Hurler syndrome, Fragile X syndrome and Prader Willi syndrome was diagnosed. This process is tried on a total of 127 photographs, 48 of them that Fragile X syndrome, 30 of Hurler syndrome, 12 of them Prader Willi syndrome and 37 of them belong to the healthy child photos. After preprocessing on images, significant points which were determined on the face depend on the type of syndrome had been identified as semi automatic. From these points features were determined, and a database of feature vectors was created. Artificial neural network was created using Matlab programme and trained by the data base, was tested using Leave One Out method so groups with syndromes and groups with healty were classified.

Important Links:

Go Back