The chord length transform and the segmentation of crossing fibres

Segmentation of crossing fibres is a complex problem of image processing. In the present paper, various solutions are presented basing on tools of morphological image processing. Two new image transforms are introduced – the lineal distance transform and the chord length transform. Both transforms are applied to two‐dimensional images and their results are three‐dimensional images. Thus, the segmentation problem originally formulated for crossing fibres observed in a two‐dimensional image can be reformulated as a segmentation problem in a three‐dimensional image. This can be solved by a segmentation in the three‐dimensional image. Algorithms for the lineal distance transform and the chord length transform are given and their use in image analysis is demonstrated. Furthermore, the chord length distribution function of the foreground of a binary image can efficiently be estimated via the chord length transform.     

Semi-automated measurement of true chord length distributions and moments by video microscopy and image analysis

The distribution of the lengths of airspace chords in pulmonary parenchyma characterizes many architectural features of the alveoli and alveolar ducts. Laborious to obtain manually, the distributions and density functions may be acquired semi-automatically by video microscopy, digitization and image processing. The accuracy of the estimation is influenced by the microscopical methods and also by the techniques used (i) to convert the digitized grey-scale picture to a two-valued image, (ii) to collect the chord lengths and (iii) to compensate for finite field widths. The last problem arises because some chords are completely visible within a field while others are only partially seen, since one of the two air-tissue boundaries lies outside the field of view. This error systematically biases the observed distribution. This paper contains solutions to hardware, software and analytic problems encountered while developing the capability to measure airspace chord length density functions semi-automatically. Formulas for estimating the true chord length density function from samples of observed chord lengths are presented. Also given are formulas for the estimation of the first and second moments of the true chord length distribution from the means of observed chord lengths. These techniques of image preparation and analysis should be suitable for characterizing particle, grain or cell size distributions, especially where many profiles fall partially outside the field of view.     

Determination of a homogeneity factor for composite materials by a microstructural image analysis method

The physical properties of particle‐reinforced composite materials are highly affected by the distribution of particles within a matrix material. In this study, a microstructural image analysis method with a new distribution index for quantifying the degree of distribution in composite materials was developed. The free‐path spacing between particles was measured to calculate the distribution (D) index based on the coefficient of variation. The proposed method was applied to six digitally created reference patterns as representative binary composite microstructures and three actual ceramic‐matrix composites, respectively. It is found that the D index increased from 0.00 to 0.67 depending on the degree of distribution or homogeneity level based on the reference patterns. The homogeneity levels for the binary composites are then classified from a perfect (maximum) to very low level (minimum) based on increasing D index values, where a high D index presents a poorer distribution. The results obtained for reference patterns and metal silicide‐refractory oxide composite microstructures indicate that the proposed method is a useful tool to quantify the degree of distribution with high accuracy, and can be efficiently used for different types of composite microstructures.     

Measuring the surface area of a cell by the method of the spatial grid with a CSLM—a demonstration

The spatial grid is a method for estimating the surface area of particles. A stack of perfectly registered sections is the essential prerequisite for its use. The confocal scanning light microscope provides such a stack by optical sectioning. The spatial grid method is briefly described and applied to an osteocyte lacuna in dry mineralized human mandible. This type of cell was chosen because of its very complex shape. The variance of the area estimate is studied and compared with the results of a simulation.     

The saucor, a new stereological tool for analysing the spatial distributions of cells, exemplified by human neocortical neurons and glial cells

The 3D spatial arrangement of particles or cells, for example glial cells, with respect to other particles or cells, for example neurons, can be characterized by the radial number density function, which expresses the number density of so-called 'secondary' particles as a function of their distance to a 'primary' particle. The present paper introduces a new stereological method, the saucor, for estimating the radial number density using thick isotropic uniform random or vertical uniform random sections. In the first estimation step, primary particles are registered in a disector. Subsequently, smaller counting windows are drawn with random orientation around every primary particle, and the positions of all secondary particles within the windows are recorded. The shape of the counting windows is designed such that a large portion of the volume close to the primary particle is examined and a smaller portion of the volume as the distance to the primary object increases. The experimenter can determine the relation between these volumina as a function of the distance by adjusting the parameters of the window graph, and thus reach a good balance between workload and obtained information. Estimation formulae based on the Horvitz-Thompson theorem are derived for both isotropic uniform random and vertical uniform random designs. The method is illustrated with an example where the radial number density of neurons and glial cells around neurons in the human neocortex is estimated using thick vertical sections for light microscopy. The results indicate that the glial cells are clustered around the neurons and the neurons have a tendency towards repulsion from each other.     

Classification of spatial textures in benign and cancerous glandular tissues by stereology and stochastic geometry using artificial neural networks

Stereology and stochastic geometry can be used as auxiliary tools for diagnostic purposes in tumour pathology. Whether first-order parameters or stochastic-geometric functions are more important for the classification of the texture of biological tissues is not known. In the present study, volume and surface area per unit reference volume, the pair correlation function and the centred quadratic contact density function of epithelium were estimated in three case series of benign and malignant lesions of glandular tissues. The information provided by the latter functions was summarized by the total absolute areas between the estimated curves and their horizontal reference lines. These areas are considered as indicators of deviation of the tissue texture from a completely uncorrelated volume process and from the Boolean model with convex grains, respectively. We used both areas and the first-order parameters for the classification of cases using artificial neural networks (ANNs). Learning vector quantization and multilayer feedforward networks with backpropagation were applied as neural paradigms. Applications included distinction between mastopathy and mammary cancer (40 cases), between benign prostatic hyperplasia and prostatic cancer (70 cases) and between chronic pancreatitis and pancreatic cancer (60 cases). The same data sets were also classified with linear discriminant analysis. The stereological estimates in combination with ANNs or discriminant analysis provided high accuracy in the classification of individual cases. The question of which category of estimator is the most informative cannot be answered globally, but must be explored empirically for each specific data set. Using learning vector quantization, better results could often be obtained than by multilayer feedforward networks with backpropagation.