Genetic Syndrome Identification: An Image Processing Approach

Abstract

Genetic syndromes often cause intellectual disability and facial dysmorphism. These dysmorphic facial characteristics provide distinct information which can be used in the identification of various syndromes. However, identification of dysmorphism using a computer is complex since it involves scientifically measuring the facial features and correlating with pathological proportions. Here we describe our work on image-based screening for dysmorphic children which uses the basic concept of symmetry and geometry of the human face. Frontal face images are used to facilitate syndrome identification. Various facial landmarks are selected, different angles and distances are computed for normal and affected face images. It was observed that Support Vector Machine with Radial Basis Function kernel (SVM-RBF), Linear Discriminant Classifier (LDC), and Quadratic Discriminant Classifier (QDC) are suitable to classify the faces and bucket them in dysmorphic and non-dysmorphic categories. Maximum classification accuracy achieved is 89.19% by Linear Discriminant Classifier.

KEYWORDS:

• Classifier
• Facial and anatomical landmark
• Feature extraction
• Syndrome

Acknowledgment

The authors would like to thank Dr A S Bhalchandra, Head of Electronics and Telecommunication Engineering Department, for her encouragement, experienced guidance and support. The authors would also like to thank Dr P B Murnal, Principal, Government College of Engineering, Aurangabad, for his advice and support.

Additional information

Notes on contributors

Archana P. Ekbote

Archana P Ekbote received her BE degree in electronics and telecommunication engineering from Dr Babasaheb Ambedkar Marathwada University, Aurangabad, MS, India in 1997 and ME degree in electronics and telecommunication engineering from University of Mumbai, MS, India in 2009. She is pursuing PhD degree from Dr Babasaheb Ambedkar Marathwada University, Aurangabad, MS, India. Currently, she is an assistant professor in the department of Information Technology at Vidyavardhini's College of Engineering & Technology, University of Mumbai, MS, India. She has published three conference papers. Her research interest includes signal and image processing, pattern recognition. Corresponding author. E-mail: [email protected]

Varsha R. Ratnaparkhe

Varsha R Ratnaparkhe received her BE degree in electronics and telecommunication engineering from Dr Babasaheb Ambedkar Marathwada University, Aurangabad, MS, India in 1989; MTech (EDT) degree from C E D T Aurangabad, MS, India in 1996 and the PhD degree in electronics from Swami Ramanand Teerth, Marathwada University, Nanded, MS, India in 2008. Currently, she is associate professor in Government Engineering College, Aurangabad, MS, India. She has published many research papers in journals (12) and conferences (8). Her area of interest includes image processing, pattern recognition. E-mail: [email protected]