Classification of Cancerous and Non Cancerous Cells in H & E Breast Cancer Images Using Structure Descriptors

Main Article Content

B. Lakshmanan, et. al.

Abstract

Computer-assisted earlier detection and diagnosis of breast cancer are possible by analyzing morphological features in histopathology images. In this article, we propose computer-aided tool for accurate detection of the abnormal cells using shapes and morphological based features extracted from the segmented region of the input image. Different segmentation techniques have been performed and their performance is compared to find out the best segmentation algorithm for further processing ie., classification. Morphological descriptor such as major and minor-axis length, cell area, perimeter and eccentricity are extracted from the segmented region of the image. By analyzing the cell mass and contour of cells in the given input image, cells are classified as cancerous and non-cancerous. The results are validated using benchmark dataset images taken from Mitosis Atypia14 grand challenge in which 180 images are taken for training and 120 images for testing. The proposed method provides improved F-score of 94.12% compared with other previously proposed frameworks.

Downloads

Download data is not yet available.

Metrics

Metrics Loading ...

Article Details

How to Cite
et. al., B. L. . (2021). Classification of Cancerous and Non Cancerous Cells in H & E Breast Cancer Images Using Structure Descriptors. Turkish Journal of Computer and Mathematics Education (TURCOMAT), 12(10), 1830–1839. Retrieved from https://turcomat.org/index.php/turkbilmat/article/view/4656
Section
Research Articles