Computer Vision Theoretic Approach for Breast Cancer Diagnosis: Commonly Perceived Diagnostic Significance of Cytological Features and Feature Usability Analysis of an Existing Breast Cancer Database
Editor(s): E. Y. K. Ng; U. Rajendra Acharya; Rangaraj M. Rangayyan; Jasjit S. Suri
Author(s): Hrushikesh Garud, Debdoot Sheet, Jyotirmoy Chatterjee, Manjunatha Mahadevappa, Ajoy Kumar Ray, Arindam Ghosh
Published: 2013
Author Affiliations +
Abstract
Cytopathology is a branch of pathology that studies and diagnoses diseases on the cellular level using samples of free cells or tissue fragments. The tissue fragments can be collected by exfoliation or by intervention techniques such as fine-needle aspiration (FNA). FNA is widely used for evaluation of a variety of breast abnormalities. It has been shown in numerous studies to be a good screening tool for diagnosis of breast lumps in symptomatic patients. Microscopic appearance of the nuclei, cells, cellular arrangement in clusters, and background elements in the smeared aspirates provide clues for evaluation and diagnosis. Several visual clues (features) relevant for the diagnosis of benign or malignant conditions of the breast abnormality as obtained via microscope have been reported in the literature and are used by experts in the decision-making process. A list of eighteen such adequate cytological features is given in Table 13.1. The list was prepared with the help of scholarly texts in clinical pathology.The list is subdivided in the four categories (1) aggregate properties, (2) background properties, (3) nuclear properties, and (4) cellular properties according to the elements whose features are used as evidence for diagnosis. Currently, cytological diagnosis of the breast lump is based on the subjective assessment of the microscopic appearance of the aspirate. As a result, difficulties in maintaining consistency and reproducibility are inevitable. A review of the literature served to highlight the following limitations of fine-needle aspiration cytology (FNAC) leading to equivocal diagnosis: (1) inadequate or nonrepresentative sampling, and (2) the overlap of cytological features of benign and malignant lesions due to the nature of the lesion. Image analysis of breast FNAC slides by computer vision techniques is helpful in overcoming some of these limitations. Incorporating the practice of FNA with an expert system embedded in a microscope aidspathologists/ cytopathologists (experts) in a speedy and accurate assessment of the slides through the use of quantitative and objective feature assessment for diagnostic decision making.
Online access to SPIE eBooks is limited to subscribing institutions.
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Breast cancer

Breast

Diagnostics

Computer vision technology

Machine vision

Databases

Fine needle aspiration

Back to Top