In many disciplines, such as biology, botany, geology, materials science and medicine, quantitative image analysis is being used to an increasing extent. In materials science this technique makes it possible to relate the microsctructure to the mechanical properties. In this review we shall show that image analysis can be applied in a fractographic study to characterize quantitatively the morphology of fracture. Such an analysis provides information which, together with that obtained by mechanical tests, enables an explanation of the mechanism of rupture to be made.

Relative abundance of tumour angiogenesis has been shown to be of clinical relevance in cancers of various locations such as the ovary. Nevertheless, several problems are encountered when quantifying tumour microvessels: (i) as many other tumour markers, vascularity pattern is often heterogeneous within the tumour mass and even within the same histological section. As a consequence, an adequate acquisition method must be developed for accurate field sampling. (ii) Manual microvessel counting is long, tedious and subject to poor reproducibility. Introduction in routine practice requires a fast, reproducible and reliable automatic image processing. In this study we present an original procedure combining a slide scanner image acquisition and a fully automatic image analysis sequence. The slide scanner offers the advantage of recording an image of the whole histological section for subsequent automatic blood vessel detection and hot spot area location. Microvessel density and surface fraction were measured for the whole section as well as within hot spots. Different immunostaining methods were tested in order to optimise the procedure. Moreover, the method proposed was submitted to a quality control procedure, with reference to interactive identification of microvessels at scanner level. This experiment showed that 93 to 97% of blood vessels were detected, according to the staining protocol used. Colour figures can be viewed on http://www.esacp.org/acp/2003/25‐2/kim.htm.

The knowledge of the mean grain size of ceramics is a very important problem to solve in the ceramic industry. Some specific methods of segmentation are presented to analyse, by an automatic way, the granulometry and morphological parameters of ceramic materials. Example presented concerns cerine materials. Such investigations lead to important information on the sintering process.

The scope of this paper is to investigate by an automatic image analysis technique the dispersion of phases in concrete. Three morphological and statistical tools were used: co-occurrence matrices, and simple and crossed-covariance. It was shown (1) that there is a repulsion between gravel, (2) that gravel and air-voids are surrounded by matrix (cement paste and sand), and (3) that the dispersion of gravel and air-voids is perfectly uniform.

The scope of this paper is to present the main tools of image analysis to investigate materials and, specially, civil engineering ones. First the acquisition methods are described. The different operators for filtering, segmentation and binary image processing are presented and illustrated on different images. The influence of the observation field on these operators and the bias correction is also introduced. Then the problem of the parametrical characterization is presented: stereological parameters and functions related to size distributions, dispersion and anisotropy. Finally, the model methods based on image analysis are recalled. Some annexes illustrate this paper to precise main basic notions to understand the morphological tools.

This paper illustrates the use of automatic image analysis technique to investigate the morphology of cement, concrete and fibre-reinforced concrete. First the methods to be used for powders and secondly for mortar and concrete are introduced. The dispersed phases are characterized by classical morphological parameters: these also enable to accede to the hydration process. The covariances give quantitative information on the homogeneity and dispersion of the different components: gravel, air-voids and cement paste. Air-voids are characterized by granulometric distributions and their mean free paths. Rose of directions gives information on feature orientation: fibres, microcracks for fibre-reinforced concrete, etc. Finally probabilistic models can be used to simulate the microstructure of such materials.

This short paper will present a two-dimensional (2D) model of concrete material, based on probabilistic models: it is a combination of a Voronoi tessellation for the gravel, followed by a Boolean model of spherical grains for the air-voids, with elimination of the air-voids-gravel intersections so that there is no contact between them. The model was tested via crossed-covariance. Results on true and simulated structures are in good agreement. This is a first step towards a 30 model. (C) 2001 Elsevier Science Ltd. All rights reserved.

This paper gives a brief over-view on the different probabilistic models, which can be used to model the microstructure of monophased or polyphased ceramics. These models require few parameters to describe their morphology. Many of them possess stereological properties and are easily tested using mathematical morphology.

In order to access to the air-void distances in concrete by automatic image analysis, three different methods were proposed, used and discussed: half-distances between two voids nearest neighbours, called count-dilation; half-distances between all neighbours, called SKIZ; and distance function. They were utilized in a euclidean and geodesic way. Some of them can replace favourably the manual methods proposed in the different standards.

Ceramics are always fabricated from powders or a mixture of powders. Bulk materials are obtained either via a liquid (slurry infiltration), solid (sintering, electromelting) or gas (chemical vapour deposition or infiltration) process routes. The ceramics can be monolithic (alumina, silicon carbide, silicon nitride, ...). or polyphased (alumina-zirconia, silicon carbide-silicon nitride ...). All these ceramics, whatever are their state - powder, green or sintered materials - can be characterized from their morphological point of view by using automatic image analysis. That is also the case for ceramic matrix composites (CMCs) or cermets (ceramic metal), which are generally obtained by sintering via a liquid phase route from a metal matrix and a dispersed ceramic phase.The knowledge of the morphological characteristics of a ceramic is presently very important as these parameters control most often the mechanical strength or electrical properties of the material. So, if is necessary to establish relationships between morphology and physical properties in order to have a return to the scientists in charge of the process to improve the ceramic material.Before any image analysis, it is necessary to observe the microstructure of the ceramics at the microscopic scale. It corresponds, in fact, to one of the most difficult problem to solve. After a classical polishing to reveal the ceramic grain boundaries, one can use generally either thermal etching (which can modify the size distribution if time and temperature are too high) or chemical etching (which do not reveal generally all the grain boundaries). More recently plasma etching technique can be used, which offers nice results specially in the case of silicon nitrides. Observations are generally made either with an optical or a scanning electron microscope.The first step in image analysis is to segment the image in order to enhance (or defect) all the grain boundaries. The different methods of segmentation which can be used in the case of porous ceramics, of ceramics with a low porosity and of bulk ceramic are presented and illustrated.At that stage, the ceramic can be considered as a two-phase system : 1) pores and ceramic; 2) matrix and refractory phase.The methods of parametrical characterization of the microstructure can be classified in two types :if the grain phase has not being segmented, classical stereological parameters can be used : the P(I) function will give access to the main morphological size parameters;2) if there is a correct segmentation of the ceramic grains, then an individual analysis of the size and of the shape can be performed, like for the powder analysis.These different cases are presented and some applications are given in the case of BaTiO3, UO2 and glass.For materials presenting an anisotropy (due to a hot-press process for example) or for ceramics with whisker or fiber reinforcement, anisotropy can be obtained in using the rose of intercepts or of directions. These methods can lead also to the size distribution of the neighbour number of a grain in R-2, which permits to inform on the grain growth.The sampling and homogeneity of ceramics can also be known in using the covariance or the regionalized variogram. That is the only way to reach such information regarding the homogeneity at the morphological scale. Examples are given in the case of UO2.Finally modelling can also be performed by automatic image analysis. Stereology is often sufficient to estimate 3D parameters, Nevertheless, the topological parameters in R-3, the phase dispersion parameters, ... are not accessible from measurements in R-2. Probabilistic models are the elegant solution. They allow, for example, to accede to the number of particles per unit volume without any hypothesis and without using a serial sectionning Two examples are presented, in the case of MOX and WC-Co materials.

For the investigation of civil engineering concrete, the hydration of the cement paste is important to know. The parameter investigated for that is the hydration degree, a. Different techniques are used to study this parameter: In this work, we have chosen to determine ct using scanning electron microscope (SEM) and automatic image analysis.On a SEM image of a cement paste, two phases can be observed : hydrated phase which appears in grey and anhydrous phase (NH) which appears in white. So, after an image processing, the surface area of the anhydrous phase can be calculated: it is equal to the volume fraction of this phase, V-V(NH). In this case, one has :alpha = 1 - weight of non hydrated cement / initial weight of anhydrous cement= 1 - rho anhydrous cement * V-V(NH)* V-test tube / initial weight of anhydrous cementFor this work, two cement pastes were prepared. They differ by their water/cement ratio (w/c), in weight, which are respectively 0.5 and 0.3.50 images per specimen were acquired by SEM using back scattering electron mode. This number requires to use an automatic process. It is made of three stages :- a filtering of the grey image,- a segmentation of the anhydrous phase,- a binary process to perform the final image.The kinetic of hydration is principally studied with the hydration degree. But it is interesting to analyse the stereological parameters, like the specific connexity (N-A(NH)) and intercept (N-L(NH)) numbers. They allow to calculate the specific surface area (S-V(NH)), the mean free path ((L) over bar(3)(NH)), the mean curvature ((H) over bar(3)(NH)), and the mean surface area ((A) over bar(NH)) of the anhydrous phase.So, we note that, for a given time the hydration degree of a cement with ratio w/c = 0.3 is lower than the one of a cement with w/c = 0.5 For w/c = 0.3, the hydration evolves very slowly rather than the hydration of the cement with w/c = 0.5 increases rapidly between 2 to 14 days, before taking an asymptotic value.To validate our results obtained by automatic image analysis, they were compared them with results obtained by thermogravimetric analysis (TGA) : the evolution of the hydration degree by TGA gives similar results as those obtained by automatic image analysis.The analysis of the stereological parameters shows that the specific surface area of the anhydrous phase decreases during the first days, and then takes an asymptotic value. It can be interpreted like an evolution of the surface energy: during the first days, there is a minimization of the superficial energy between phases, then this interface takes a stationary value.The evolution of ((L) over bar(3)(NH) and (H) over bar(3)(NH) of the anhydrous phase shows that the smaller anhydres disappear in the first days, and after their mean size stays constant.Finally the study of (A) over bar(NH)/((L) over bar(3)(2)(NH) shows that, after 4 days, the shape of the anhydres stays constant: hydration is an isotropic phenomenon.

OBJECTIVE To assess the number of nuclei required for significant image cytometry DNA ploidy measurements on one archival case of breast cancer. STUDY DESIGN From one case of aneuploid DNA breast cancer, 18 subsets made up of 152-1,524 for the whole population of undamaged nuclei and made up of 74-735 epithelial nuclei had DNA measured. DNA ploidy type and five DNA ploidy indices, allowing DNA ploidy histogram interpretation were evaluated on each population. RESULTS Three hundred nuclei were always sufficient for DNA typing, whereas reliable results for DNA ploidy indices required at least 750 nuclei. CONCLUSION To DNA measure the above number of nuclei, fully automated image cytometry DNA ploidy measurements are required.

The aim of the present study is to propose alternative automatic methods to time consuming interactive sorting of elements for DNA ploidy measurements. One archival brain tumour and two archival breast carcinoma were studied, corresponding to 7120 elements (3764 nuclei, 3356 debris and aggregates). Three automatic classification methods were tested to eliminate debris and aggregates from DNA ploidy measurements (mathematical morphology (MM), multiparametric analysis (MA) and neural network (NN)). Performances were evaluated by reference to interactive sorting. The results obtained for the three methods concerning the percentage of debris and aggregates automatically removed reach 63, 75 and 85% for MM, MA and NN methods, respectively, with false positive rates of 6, 21 and 25%. Information about DNA ploidy abnormalities were globally preserved after automatic elimination of debris and aggregates by MM and MA methods as opposed to NN method, showing that automatic classification methods can offer alternatives to tedious interactive elimination of debris and aggregates, for DNA ploidy measurements of archival tumours.

Automatic image analysis is an efficient tool to quantify the morphology of materials. Moreover, it can aid to understand their mechanical behaviour. Several applications of automatic methods are presented to investigate concrete reinforced by ribbon shaped amorphous cast iron fibres. Introducing ribbons into the plain matrix entrapped air voids. This affected the workability and, later on, the compressive strength of the fibre reinforced concrete (FRC). Both were improved by additions of superplasticizer in order to keep the water to cement ratio constant. The influence of the superplasticizer and fibre contents on the compactness of the FRC was characterized by the dimensional and the spatial distributions of the air voids. The orientations of fibres and microcracks were quantified by Fourier image transforms. Due to the casting procedure of the FRC, the fibres were located in “horizontal layers”, perpendicular to the casting axis. Whatever the direction of compression with respect to the layers of fibres, the microcrack network was getting more and more oriented in the direction of compression as stresses increased. The analysis of fibre and microcrack orientations suggests that, under uniaxial compression, the inelastic strain domain should be characterized by an anisotropic biaxial damage model, whose principal axes are the orthogonal and parallel directions to the layers of fibres.

In this study, ribbon‐shaped amorphous cast‐iron fibres were used to reinforce a concrete matrix. X‐ray photographs have been taken to detect fibres in situ. Their orientation has been investigated by automatic image analysis methods. However, this measurement should not be influenced by the digitization on the square frame of the analyser. For that purpose, the Fourier transform was used rather than the rose of direction method. This analysis revealed the transverse isotropic nature of the spatial arrangement of these fibres, whose axis of revolution corresponds to the concrete casting axis. Such a morphological characterization of the fibre‐reinforced concrete reveals its mechanical behaviour.

OBJECTIVE:To investigate, for seven samples, the influence of unwanted elements (e.g., remains of erythrocyte cell membranes, sliced nuclei, damaged nuclei and aggregates) on image cytometry DNA ploidy measurements.STUDY DESIGN:Two normal reference tissues (brain and breast), one breast cancer and four brain tumors were studied. For each sample, the influence of the different classes of debris on DNA ploidy histograms and indices was evaluated.RESULTS:The influence differs regarding each class of debris and the index to be evaluated.CONCLUSION:Strict and precise elimination of debris and aggregates is required. Moreover, strategies and efforts that must be applied to automated elimination of these unwanted elements must be a direct function of the bias they introduce into DNA ploidy measurements.

The purpose of this paper is to present the results obtained using probabilistic models (Boolean model, dead leaves model and mosaic model particularly) for the description of the coarsening of the grains in two-phased refractory compounds. SEM (Scanning Electron Microscope) micrographs of the samples (WC-Co, TiC-Co and W-(Ni,Fe) systems) were analyzed and several models were tested on each of these materials. The Choquet capacity and different compact sets were then used in order to determine the best model which describes the macrostructure of the grains. Afterwards the parameters of the model allow to characterize the kinetics of the growth of the materials. It was determined that the WC-Co cermets were described by a Boolean model with Poisson polyhedra primary grains, whatever the degree of sintering was. In the case of materials with rounded grains, it seems that a single model is not sufficient in order to describe the whole process of grain coarsening.

La majorite des tris d'objets biologiques effectues en analyse d'images fait appel a la mesure de parametres (objet par objet) suivie d'une analyse des donnees. La morphologie mathematique offre des outils susceptibles de traiter les images globalement. Jusqu'a present, cette opportunite a ete sous employee pour les tris d'objets. Le but du present article est de proposer des methodes globales de tri, appliquees a l'elimination des agregats et des petits debris, pour la mesure de l'ADN ploidie des tumeurs solides. Les resultats illustrent la puissance et l'efficacite des traitements globaux par morphologie mathematique utilises comme methode de tri.

In the case of absence of overlaps, roughness investigation can be studied from IR2 x IR function analysis methods. These IR2 x IR functions correspond to a true topography or a perspective representation. They can be quantitatively described directly from basic stereological parameters or after morphological transformations. Some examples in material science are given using scanning electron or confocal microscope images.

The influence of microstructural features of cutting tools in tungsten carbide-cobalt cermets (WC-Co) on their mechanical behaviour is well established. The study of these features needs to segment WC grains in the cobalt matrix. This segmentation had been manually performed for a long time, but the development of new image processing tools allow now to automatically achieve a good segmentation. A method of WC-Co segmentation is introduced and the results obtained on this one are compared to the results obtained on a manual segmentation in order to validate the segmentation algorithm.