Certain investigations on prenatal down syndrome detection using soft computing and evaluat

Abstract

Down syndrome (otherwise called as trisomy 21) is the disorder of the fetus which can be identified by examining the ultra sonogram images at first and second trimester stages. This can be occurred at the ratio of 1:800 live births in world. The integration of extra chromosome in 23 human chromosome leads to genetic disorder in fetus, which is the main cause of DS. The maternal duration and the morphological analysis of Nuchal Translucency (NT) in fetus are determined by first and second trimester stages. The DS can be screened by analyzing the trimester stage images manually which achieves 85% of average detection rate with 5% to 10% error rate with respect to different radiologist. The nasal bone (small bifid structure in fetus) forms in the fetus after first trimester stage of the fetus. The first trimester images can be obtained between 11-14 weeks of the fetus development. There is no nasal bone formed in first trimester stage. The second trimester images can be obtained between 15-20 weeks of the fetus development with the growth of nasal bone. The radiologist determines the length of the nasal bone at second trimester stage for DS detection at an earlier stage as by manual. Due to this limitation, the researchers are concentrated their research on impact of nasal bone length in second trimester stages for an efficient detection of DS. The Genetic Disorder of fetus leads to the formation of Down Syndrome (DS) which can be screened manually by screening the first and second trimester ultra sonogram images. This can be fully automated with the help of computer aided approaches proposed in this research work. The DS can be screened by enhancing the fetus image using Adaptive histogram equalization (AHE) technique. Then, Gabor multi resolution transform is applied on the enhanced fetus image in order to convert the spatial domain fetusimage into multi resolution fetus image. The features as Effective Binary Pattern (EBP), Grey Level Occurrence Matrix (GLCM) and Local Derivative Pattern (LDP) are extracted from the enhanced Gabor transformed fetus image and then these features are trained and classified using Adaptive Neural Fuzzy Inference System (ANFIS) classifier, which classifies the fetus image into either normal or abnormal. Further, morphological based segmentation technique is applied on the abnormal classified fetus image to segment the nasal bone region. The segmented nasal bone region is compared with clinical diagnosis results to detect DS in fetus image. newline