Automated classification of ovarian tumors using GLCM and Tamura features with Light GBM classifier
Smital Dhanraj Patil, Pramod Jagan Deore, Vaishali Bhagwat Patil · 2023
Ovarian abnormalities, such as ovarian cysts, tumours, and polycystic ovary (PCO) are one of the most serious concerns for women’s health after breast cancer. In females, ovarian cysts must be accurately diagnosed to make informed decisions at the appropriate time. Classification and identification of ovarian masses are quite complicated due to complexity and resolution of image background. So features of such images play a vital role in automatic classification. This paper aims to implement the classification of ovarian abnormalities using GLCM and Tamura texture features. Various texture features like coarseness, contrast, directionality, and roughness are extracted and Light GBM classifier is used to achieve the classification. Performance measures like F1-score, precision, accuracy, and recall are used to analyse the results of classification. The classification accuracy for combination of GLCM and Tamura texture features is found to be 72% for proposed work without segmentation.