Cancer Cell Image Analysis and Visualization

Cancer Cell Image Analysis and Visualization

Tae-Yun Kim (Inje University, Korea), Hae-Gil Hwang (Inje University, Korea) and Heung-Kook Choi (Inje University, Korea)
Copyright: © 2012 |Pages: 11
DOI: 10.4018/978-1-4666-0909-9.ch018
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

We review computerized cancer cell image analysis and visualization research over the past 30 years. Image acquisition, feature extraction, classification, and visualization from two-dimensional to three-dimensional image algorithms are introduced with case studies of bladder, prostate, breast, and renal carcinomas.
Chapter Preview
Top

Materials And Methods

Image Acquisition

Early microscopic image acquisition involved attaching an analog charge-coupled device (CCD) camera to a microscope (Choi, 1994). Then, the analog signals were converted into digital signals using a frame gabber. Recently, digital CCD cameras have been attached directly to microscopes.

Histological Cell Feature Extraction

Once cells are stained with a biochemical reagent, the most important process is to identify the cell nucleus, which involves subtracting the region of interest (ROI) from the background. Then, we quantify the features of the cell nucleus, including the size, perimeter, major and minor axes, and cell numbers. Textural features include entropy, contrast, moment, and intensity variation. We can calculate more than 1,000 features. Then, we select significant features using statistical analyses (Cox, 1972).

Finally, we determine the cell densitometry of an area (Choi, 2007; Kayser, 1992). Generally, the most important objects in a ROI are the cell nuclei, for which we have calculated the cell features. Figure 1 shows the (a) morphology, (b) texture, (c) intensity, and (d) special staining for molecules in sample cells.

Figure 1.

Examples of various cell characteristics

Data Classification

Normally, two methods are used for numerical data classification: statistical classification (i.e., the conventional method) and neural network methods. Statistical analyses include multivariate, factor, regression, principal components, and independent component analyses (John, 1992). Neural networks methods include back propagation networks, Hopfield networks, and adaptive resonance theory (Choi, 2007). Other classification methods are fuzzy theory and supported vector machines. The best method depends on the characteristics of a particular image.

Complete Chapter List

Search this Book:
Reset