New quantitative features for the morphological differentiation of abnormal lymphoid cell images from peripheral blood

Abstract

Image analysis may assist in quantifying cell morphology turning qualitative data into quantitative values. New cytological variables were established based on geometric and colour/texture features to contribute to a more accurate and objective morphological assessment of lymphoid cells and their association with flow cytometry methods may be interesting to explore in the next future.

Optimizing morphology through blood cell image analysis

“…The vertical axis gives the number of pixels corresponding to the different intensity intervals. From each histogram, 6 classical first order statistical features are usually calculated as follows: mean, standard deviation (SD), skewness, kurtosis, en ergy (uniformity), and entropy (variability) …”

“…As an example of statistical features calculated from the histogram, Figure (see) shows 2 original images corresponding to Sézary cells acquired by the CellaVision DM96 (DM96). The 2‐axis plot shows the histograms obtained from the magenta component in the grayscale images, in which the pixel count levels were obtained using the scientific software matlab (produced by MathWorks, MA, USA).…”

“…Over both curves, it is possible to obtain texture features by calculating parameters such as mean, standard deviation, skewness (measure of asymmetry), and kurtosis (measure of uniformity). Reference presented a study in which the granulometry was applied for the morphological differentiation of abnormal lymphoid cells.…”

Image processing and machine learning in the morphological analysis of blood cells

“…Color is also closely related to the visual cell appearance. Besides RGB, different authors use other color models to obtain quantitative features . The overall idea is to take advantage of the complementary information supplied by alternative colors (cyan, magenta, and yellow in the CMYK model) and attributes such as hue, saturation, and brightness (HSV model) or lightness and chromaticity (in Lab and Luv models).…”

“…Most recent studies using GLCM have been performed mainly for differentiating among normal leukocytes and blast lymphoid cells as well as in bone marrow images to distinguish erythrocyte precursor cells stages . However, recent advances have been reported in the definition of texture features using both granulometric and morphological texture features able to discriminate between a wide range of ALCs, blasts, and RL …”