Author/Editor		Lavrenčak, Jaka
Title		Pomen svetlih področij v celičnih jedrih obarvanih po Feulgnovi metodi
Type		monografija
Place		Ljubljana
Publisher		Medicinska fakulteta
Publication year		1998
Volume		str. 69
Language		slo
Abstract		Light microscopy examination provides the basic information about nuclear morphology and chromatin distribution pattern. Image cytometry is a method for objective and quantitative analysis of chromatin distribution and organisation. Features describing those facts are nuclear texture features. Round, bright areas (BA) have often been seen in images of nuclei that were stained with Feulgen-thionin method and scanned by image cytometer. The size, number and distribution of BA are not constant. In our study, we attempted to answer two questions; (i) do the BA have any influence on the nuclear texture features and (ii) what is the percentage of nuclei with the BA that can influence the results of nuclear texture features analysis of a particular sample. The cell samples of HL6O cell line, the smears of normal human vocal cords and the smears of breast carcinoma were taken for image cytometry analysis. All the slides were air-dried and stained by Feulgen-thionin method. Image analysis was performed by Cyto-SavantTM (Oncometrics Imaging Corp., Vancouver, Canada). The images of nuclei were classified into two groups: nuclei with the BA (group A) and nuclei without the BA (group B). Probability density functions of nuclear texture features and matched pairs t-test were used for statistical analysis. The new algorithm was applied for detection and description of the BA. The simulation model was used to analyse what is the percentage of nuclei with the BA that can influence the results of nuclear texture features analysis. The BA influenced nuclear texture features analysis. Nuclear texture features with the best discriminative power between the groups A and B for all three cell samples were, (a) range extreme, (b) fractal area and (c) run length. Their discriminative ability ranged between 64% and 84%. (Abstract truncated at 2000 characters).
Descriptors		HL-60 CELLS VOCAL CORDS BREAST NEOPLASMS IMAGE CYTOMETRY CELL NUCLEUS CHROMATIN CEREBELLUM STAINING