Nuclear pattern recognition by two-parameter texture analysis

Label-free classification of cells based on supervised machine learning of subcellular structures

It is demonstrated that cells can be classified by pattern recognition of the subcellular structure of non-stained live cells, and the pattern recognition was performed by machine learning. Human white blood cells and five types of cancer cell lines were imaged by quantitative phase microscopy, which provides morphological information without staining quantitatively in terms of optical thickness of cells. Subcellular features were then extracted from the obtained images as training data sets for the machine learning. The built classifier successfully classified WBCs from cell lines (area under ROC curve = 0.996). This label-free, non-cytotoxic cell classification based on the subcellular structure of QPM images has the potential to serve as an automated diagnosis of single cells.

Segmentation of kidney from ultrasound B-mode images with texture-based classification

The segmentation of anatomical structures from sonograms can help physicians evaluate organ morphology and realize quantitative measurement. It is an important but difficult issue in medical image analysis. In this paper, we propose a new method based on Laws' microtexture energies and maximum a posteriori (MAP) estimation to construct a probabilistic deformable model for kidney segmentation. First, using texture image features and MAP estimation, we classify each image pixel as inside or outside the boundary. Then, we design a deformable model to locate the actual boundary and maintain the smooth nature of the organ. Using gradient information subject to a smoothness constraint, the optimal contour is obtained by the dynamic programming technique. Experiments on different datasets are described. We find this method to be an effective approach.

Markov optical texture parameters as prognostic indicators in ovarian carcinoma

Texture is a descriptive property of a surface describing the morphometric heterogeneity of complex structures. Computer aided image analysis allows optical texture measurement and analysis of gray-scale images. The authors, utilizing image analysis, prospectively studied Markov nuclear texture features to determine their relevance as prognostic indicators of survival in patients with epithelial ovarian carcinoma. Ninety-nine consecutive patients with ovarian cancer, treated initially with surgery were evaluated for their length of survival, level of cytoreduction, FIGO stage, grade, histology, and DNA index, as well as 20 Markov texture features. Markov nuclear texture features were quantified using image analysis. Mean follow-up for the study population was 64 months (median 59) with a range from 51 to 89 months. Five optical texture features showed significant correlation with length of survival. Difference entropy (P = 0.033) and information measure A (P = 0.041) were both indirectly correlated with survival while information measure B (P = 0.030), correlation coefficient (P = 0.045), and the maximum correlation coefficient (P = 0.041) were directly correlated. Only sum entropy (P = 0.035), FIGO stage (P = 0.0031), and level of cytoreduction (P < 0.0001) were independent predictors of survival in this population. Optical texture can be quantified by image analysis. Utilizing multivariate analysis, the Markov texture feature, sum entropy, was demonstrated to be an independent prognostic indicator of survival in patients with epithelial ovarian cancer. FIGO stage and optimal cytoreduction also were independent prognostic indicators of survival.

Chromatin pattern by variogram analysis

Many cytological processes such as cell proliferation, differentiation, transformation, apoptosis, etc., are accompanied by specific chromatin changes, usually identified on the basis of the relative content of euchromatin and heterochromatin. In order to achieve a quantitative, non-subjective evaluation of the chromatin pattern, two different approaches may be undertaken, one consisting in the analysis of the several morphological features of chromatin grains (size, shape, density, arrangement, and distribution), and the second consisting in the analysis of the chromatin globally considered as a coherent texture. Although the second approach appears to be simpler and more suitable, methods of texture analysis--including those specifically designed for the analysis of the chromatin pattern--are rarely applied due mainly to the unsuitability of sampling procedures and the excessive crypticism of results. As an alternative to traditional texture analysis, we suggest a method supported by a sound mathematical theory and approximately 30 years of applications in the field of geostatistics. The method, called variogram, analyzes the intrinsic structure of data sampled at different distance intervals and directions, and outputs easily understandable results. Recently, variogram analysis has successfully been exported from geostatistics to other fields (for example, ecology and epidemiology) that make use of spatially referenced variables. Based on the fact that pixels represent a perfect array of data ordered at regular distance intervals and directions, the variogram can be adopted to explore nuclear images and recognize chromatin patterns. Variograms of different nuclei can be summarized by multivariate methods without the need of previous standardization of data. This allows comparison and discrimination of chromatin patterns from mixed cell populations. Preliminary data obtained from young neurons undergoing massive apoptosis reveal a self-consistent map of nuclear changes correlated to the degenerative process.