Detection of Fetal Abnormality Using ANN Techniques
Authors :- V Rawat, V Shrimali, A Jain, A Rawat
Publication :- Biomedical Signal and Image Processing with Artificial Intelligence. EAI/Springer Innovations in Communication and Computing. Springer, 2023
This chapter proposes an approach of neural modeling for the diagnosis of fetus abnormality using ultrasound (US) images. The proposed method is a hybrid approach of image processing methods and artificial neural network as a classifier to extract fetus abnormality. In the first step, 50 US images in the DICOM format are taken from the radiologist, preprocess these images by segmenting the fetal biometric parameters using the morphological operator and the gradient vector flow algorithm. The segmented parameters were labeled as normal and abnormal fetal parameters. The extracted parameters were used to train a feed-forward back-propagation neural network. Then, in the next step, 500 US images are taken to process the neural model. The feed-forward back-propagation neural network using Levenberg–Marquardt, Bayesian Regularization, and Scaled Conjugate Gradient algorithms are analyzed and used for diagnosis and classification of fetal growth. Performances of these methods are compared and evaluated based on desired output and mean square error. Results found from the Bayesian-based neural networks are in closed confirmation with the real time results.