Classification of fetal heart ultrasound images for the detection of chd

Congenital Heart Disease (CHD) is the major cause of infant mortality accounting to about 28% of all congenital defects, thereby emphasizing the need for its early detection. Ultrasound (US) imaging modality is widely used in prenatal screening for monitoring the growth of the fetus. Clinically determining the abnormality is tedious and time consuming and depends on the expertise of the radiologist. Automated recognition of CHD from ultrasonic 2D imaging is proposed in this study which makes use of statistical features and pattern classifier such as support vector machine. Cineloop sequences with different abnormalities have been used for training. Features are classified using fine Gaussian and medium Gaussian kernels of SVM classifier with tenfold cross-validation resulting in an accuracy of 90% and 91.25%, respectively. The sensitivity obtained during the simulation is 90% and 92.5% for fine Gaussian and medium Gaussian, respectively, while the specificity is 90% for both the classifiers.