Down syndrome diagnosis based on Gabor Wavelet Transform.

Area of Science:

• Biomedical Engineering
• Computer Science
• Genetics

Background:

• Down syndrome is a genetic disorder characterized by distinct facial features.
• Accurate facial recognition is crucial for identifying individuals with Down syndrome.
• Existing methods may lack efficiency or accuracy in distinguishing Down syndrome phenotypes.

Purpose of the Study:

• To develop and evaluate a novel automated system for Down syndrome recognition using facial images.
• To explore the effectiveness of Gabor Wavelet Transform (GWT) for feature extraction in Down syndrome faces.
• To compare the performance of k-nearest neighbor (kNN) and Support Vector Machine (SVM) classifiers for this task.

Main Methods:

• Utilized Gabor Wavelet Transform (GWT) for facial feature extraction.
• Applied Principal Component Analysis (PCA) for dimension reduction.
• Employed Linear Discriminant Analysis (LDA) for further feature selection.
• Implemented k-nearest neighbor (kNN) and Support Vector Machine (SVM) for classification.
• Performed feature selection prior to PCA based on feature vector component correlation.

Main Results:

• Achieved high classification accuracies: 96% with kNN and 97.34% with SVM.
• Demonstrated the effectiveness of GWT, PCA, and LDA in capturing distinctive Down syndrome facial features.
• Identified Euclidean distance metric for kNN and linear kernel for SVM as optimal parameters.

Conclusions:

• The developed system provides an efficient and accurate method for Down syndrome recognition.
• The proposed feature selection strategy enhances the performance of the classification models.
• This automated approach holds potential for aiding in the early identification and support for individuals with Down syndrome.