Blighted Ovum detection using convolutional neural network
Feni Andriani, Iffatul Mardhiyah · AIP conference proceedings · 2019
Blighted Ovum is a state of conception that contains no fetus. Early detection of Blighted Ovum may reduce the risk of miscarriage. One tool that can be used in the detection of Blighted Ovum is by using ultrasonography. However, the detection of Blighted Ovum through ultrasound image is still difficult, because it is still very dependent on the level of knowledge and subjectivity of medical experts. One of the most successful ultrasound image detection or classification methods is Machine Learning. One of the best machine learning methods in terms of image classification is called Convolutional Neural Network method. This method consists of three stages. The first stage is the feature extraction of ultrasound image data. The second stage is the learning phase or training by using feedforward and backpropagation methods. The third stage is the phase of image classification using feedforward method. This research will develop an automatic classification algorithm on the ultrasound examination result using Convolutional Neural Network in Blighted Ovum detection. This study is also expected to assist medical experts in providing quick decisions on whether or not Blighted Ovum is present, so it can lead to prevention and rescue actions against the undeveloped fetuses. The accuracy of the algorithm implementation is less than 60%. This would happen because of the lack of data, and the difficulties in classifying the BO data from the non BO, because of its similarities.