Akademik Veri Yönetim Sistemi
International Journal of Multidisciplinary Studies and Innovative Technologies , cilt.7, sa.2, ss.84-89, 2023 (Hakemli Dergi)
This study aims to analyze and classify histopathological images using a rat model of diabetes to examine the effects of diabetes
on lung tissue. At the beginning of the study, control and diabetic groups were established using Streptozotocin (STZ). Caspase
immunohistochemical staining was used to examine changes in lung tissue. Features such as Local Binary Patterns (LBP) and
Gray-Level Co-Occurrence Matrix (GLCM) were extracted from the images. These features were analyzed to assess the impact
of diabetes on lung tissue. Subsequently, the most important features were selected and used with the Lasso method. The obtained
features were classified using four different methods: Support Vector Machine (SVM), K-nearest neighbors (KNN), Artificial
Neural Networks (ANN), and Decision Tree (DT). These methods were used for image classification, and classification results were obtained. The best classification performance was achieved with images obtained from the red and blue channels, with
accuracy rates of %91.08 and %93.87 using the ANN classifier, while images obtained from the green channel yielded the highest
accuracy rate of %87.15 with the SVM classifier. According to these results, it is evident that a classification model comprising
LBP, GLCM features, and machine learning algorithms has significant potential for objectively assessing the impact of diabetes
on lung tissues through histopathological image analysis.