Endometriosis Labelling using Machine learning
S Visalaxi, T. Sudalaimuthu, Sowmya. V.J · 2023
Endometriosis is a disease affecting the women's reproductive system. The lesion-like structure that exists in the women's reproductive organ creates an impact on their fertility. The diagnostic approach of endometriosis was performed by a radiologist using scanning procedures. Those procedures predict the occurrence of endometriosis but not the severity of endometriosis. Amidst the radiologist, machine learning techniques play a predominant role to identify the severity of endometriosis. Among all machine learning techniques, the proposed approach uses a support vector machine. Support vector machine is a contemporary technique for predicting clinical-based data. Support vector machine analyzes the influencing factor that incorporates Adnexal mass, Tube blockage, Lesion size, and lesion color for predicting the severity of endometriosis as well as classifying the endometriosis as Ovarian and Deep Infiltrating endometriosis. The execution was performed and trained accuracy obtained was 85%, test accuracy was 84.5% for radius basis function (rbf) kernel and the cross-validation score was 82.5%. Also, the available data was trained using Random Forest and Linear Regression. Among all three models, the Support vector machine outperforms well with hyper parameter as rbf for the given data to classify the endometriosis and identify the severity of endometriosis.