Amniotic Fluids Classification Using Combination of Rules-Based and Random Forest Algorithm

Abstract

One of the studies of fetal anatomy is to measure the volume and echogenicity of the amniotic fluid. This study categorizes amniotic fluid into six, such as Oligohydramnion Clear, Oligodramnion Echogenic, Polygohydramnion Clear, and Polygohydramnion Echogenic, as well as Normal Clear and Normal Echogenic. Meanwhile, the current condition in determining the category of amniotic remains a perception difference among doctors, especially in identifying volume and echogenicity study, which is always conducted manually and visually. Therefore, this research proposed a model for the classification of amniotic fluid by combining the rule-based of the Single Deep Pocket (SDP) method and the Random Forest algorithm. The rule-based used was based on the feature value obtained by extracting the Single Deep Pocket (SDP) feature. Also, the Random Forest algorithm was formed to classify amniotic fluid based on the condition of echogenicity, which includes clear and echogenic based on texture features using First Order Statistical (FOS) and Gray Level Co-occurrence Matrix (GLCM) methods. The average value performance of the proposed model showed an accuracy of 90.52%, a precision of 95.72%, a recall of 75.57%, and an F-measure of 81.51%. Considering this result, the proposed model showed an average increase in accuracy performance of 9.12%, precision of 14.92%, and recall of 0.51% value of the model in previous studies.