PCA AND LDA METHOD WITH NEURAL NETWORK FOR PRIMARY DIAGNOSIS OF GENETIC SYNDROME

Abstract: Dysmorphic diseases are the result of the congenital malformation and abnormal facial features. These abnormal facial features were used to diagnose the disease before cyto or molecular techniques are explored. A high degree of experience and expertise are essential for the identification of such diseases that includes Cleft palate, anencephaly and agenesis of limb etc. The correct genotype-phenotype correlation is a labour intensive especially for rare diseases. As the face development is controlled by several genes, it provides important hints for initial identification of genetic anomalies. The use of a computer based pre-diagnosis system can offer effective decision support for the less explored field like dismorphic diseases. Certain mathematical algorithms are there which can be used to analyse the facial features that diagnose the syndrome accurately. In this work we develop and demonstrate that accurate classification of dysmorphic faces is feasible by image processing of two dimensional face images. We test the proposed system on patient image data by constructing a dataset of dysmorphic faces published in scholarly journals, hence having accurate diagnostic information about the syndrome. Our methodology represents facial image data in terms of principal component analysis (PCA) and Linear discriminant analysis (LDA) and classification is done using Neural network. This method has been tested with 03 syndromes with few images per syndrome. A diagnosis success rate of 87% has been established. It can be concluded that a great number of syndromes indicating a characteristic pattern of facial anomalies can be typically diagnosed by employing computer-assisted machine learning algorithms.

Keywords: Principle component analysis (PCA), Linear discriminant analysis(LDA); Neural network(NN).