Quantification of Pleomorphic Nucleus in Breast Cancer

Thim, Hiew Thoong (2023) Quantification of Pleomorphic Nucleus in Breast Cancer. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text Full Text -Thim Hiew Thoong.pdf Restricted to Registered users only Download (20MB)

Abstract

Breast cancer is proclaimed as the leading mortality (cancer) with 684996 deaths (15.5% of all cancer cases amongst women) reported across the world in 2020. Awareness of breast cancer for accurate diagnostic, prognostic, as well as proper treatment planning is important. According to the Nottingham Histological Grading (NHG) system, the pleomorphic nucleus is an important component contributing to the over grade for breast cancer. To date, in-depth knowledge in quantifying pleomorphic nucleus in breast carcinoma is very limited, thus, investigation and analysis of pleomorphic nucleus features are essential. In this project, software, namely CellProfiler is used to segment the nuclei cell on the histological breast cancer image, and then extract the features of the segmented cell. While the other software, which is MATLAB is used to quantify the segmented cell with the extracted features. Validation of data is an extra step to validate the quantification results by using a Support Vector Machine (SVM) model. Based on the outcome, the final range to quantify the pleomorphic nucleus is determined as 52 to 197 for Score 1 nucleus, 222 to 316 for Score 2 nucleus; 333 to 569 for Score 3 nucleus. The data validation from the SVM model is generated Receiver Operating Characteristic (ROC) curve and this is having score 1 of for the area under the curve (AUC). Also, the scatter plot shows the pattern of the points is in a positive and strong correlation. Both the ROC and scatter plot show that the classifier is having a good performance in which it can distribute the pleomorphic nucleus into Score 1, Score 2 and Score 3 precisely according to the qualitative description as in the NHG system. Hence, this verifies that the design methodology of the quantification of the pleomorphic nucleus of breast cancer is successfully done in the study. The novelty of this study lies in its approach to quantifying the pleomorphic nucleus of breast cancer using a combination of two software tools, CellProfiler and MATLAB, and then validating the results using an SVM model. This automated approach is more efficient than manual quantification, which is time-consuming and prone to inter-observer variability. The use of CellProfiler and MATLAB enables the extraction of multiple features of the segmented cells, which can provide more comprehensive information on the pleomorphic nucleus. The validation step using an SVM model ensures the accuracy and reliability of the quantification results. The quantification ranges for each score of the pleomorphic nucleus can potentially aid in clinical decision-making, such as determining the appropriate treatment plan for breast cancer patients. Overall, this study offers a promising approach for the accurate quantification of the pleomorphic nucleus in breast cancer, which can contribute to improving the diagnosis and treatment of this prevalent cancer type. III