Ashour Amira S, Guo Yanhui, Hawas Ahmed Refaat, Xu Guan
1Department of Electronics and Electrical Communication Engineering, Faculty of Engineering, Tanta University, Tanta, Egypt.
2Department of Computer Science, University of Illinois at Springfield, Springfield, IL USA.
Health Inf Sci Syst. 2018 Nov 1;6(1):21. doi: 10.1007/s13755-018-0059-8. eCollection 2018 Dec.
Schistosomiasis is one of the dangerous parasitic diseases that affect the liver tissues leading to liver fibrosis. Such disease has several levels, which indicate the degree of fibrosis severity. To assess the fibrosis level for diagnosis and treatment, the microscopic images of the liver tissues were examined at their different stages. In the present work, an automated staging method is proposed to classify the statistical extracted features from each fibrosis stage using an ensemble classifier, namely the subspace ensemble using linear discriminant learning scheme. The performance of the subspace/discriminant ensemble classifier was compared to other ensemble combinations, namely the boosted/trees ensemble, bagged/trees ensemble, subspace/KNN ensemble, and the RUSBoosted/trees ensemble. The simulation results established the superiority of the proposed subspace/discriminant ensemble with 90% accuracy compared to the other ensemble classifiers.
血吸虫病是一种影响肝脏组织并导致肝纤维化的危险寄生虫病。这种疾病有几个阶段，这些阶段表明了纤维化的严重程度。为了评估纤维化程度以进行诊断和治疗，对肝脏组织在不同阶段的微观图像进行了检查。在本研究中，提出了一种自动分期方法，使用集成分类器，即采用线性判别学习方案的子空间集成，对从每个纤维化阶段统计提取的特征进行分类。将子空间/判别集成分类器的性能与其他集成组合进行了比较，即增强/树集成、袋装/树集成、子空间/KNN集成和RUSBoosted/树集成。仿真结果表明，与其他集成分类器相比，所提出的子空间/判别集成具有90%的准确率，具有优越性。
