Textural-Based Discriminant Analysis for Tubule Measurement in Breast Histopathology Images
Ji Xuan Chai, Xiao Jian Tan, Joseph Jiun Wen Siet, Khairul Shakir Ab Rahman, Wai Loon Cheor, Wai Zhe Leow · 2024
Tubule formation, one of the core factors determining the overall grade of breast cancer, relies upon the experience of histopathologists on manual inspection with nuanced decision-making. Tubule formation offers a unique context for grading decisions given the qualitative description of this factor in accordance with the Nottingham Histopathology Grading (NHG) system. To assess tubule formation, the tubule and tumor regions must first be identified. Here, the main focus lies within tubule measurement, introducing a textural-based discriminant approach for measuring tubules in breast histopathology images. This study explores the potential of using textural features, such as the arrangement of surrounding neoplastic cells and central lumens, as discriminant features for image classification. To organize these features into a cohesive model, statistically significant textural features were incorporated through mathematical modeling, resulting in a novel feature capable of distinguishing between tubules and non-tubule structures in breast histopathology images. To assess the effectiveness of the proposed method, a Support Vector Machine (SVM) with radial basis function (RBF) was employed. The proposed measurement method yielded promising outcomes with 0.9625, 1.0000, 0.9302, and 0.9638 for accuracy, precision, recall, and F1-score, respectively.