An image quality measure based on information theory in digital microscopy is presented. In contrast to ad hoc definitions, the approach discussed here associates image quality with information, based on the definition of entropy and channel capacity. The method takes into account the received image, the transfer characteristic of the system, and an estimate of the noise. Applications that motivate and exploit such measures are sequences of images that resemble the same object but have been recorded or processed differently. The evaluation of eight microscopic series is given as an example.
An image quality measure based on information theory in digital microscopy is presented. In contrast to ad hoc definitions, the approach discussed here associates image quality with information, based on the definition of entropy and channel capacity. The method takes into account the received image, the transfer characteristic of the system, and an estimate of the noise. Applications that motivate and exploit such measures are sequences of images that resemble the same object but have been recorded or processed differently. The evaluation of eight microscopic series is given as an example.
Nondiffracting cos beams may be used in the object space of an optical microscope for causing a nonuniform illumination. This irradiance distribution consists in a set of equidistant plane maxima, and therefore the light radiated by the sample decays in its neighborhood. We propose to observe over an object plane coinciding with one of these illumination peaks, which results in a superresolving axial effect. For that purpose, illumination and detection should be oblique processes, and a computer-assisted z-scanning process is needed in order to access the axial structure of a thick object.
Every imaging technique usually suffers more or less from two inaccuracies: nonhomogenous sensitivity of the recording device over the images area and spatial distortion. This work presents a method that corrects both errors by evaluating the changes of the images due to two small sample position shifts. The algorithm calculates a vector field that is used to determine the “undistorted” position of any point of the image and to determine the apparatus-caused inhomogeneities of the imaged intensities leading to an image that is free from both kinds of imaging errors. In this paper, the theoretical basis of the algorithm, the correction of a test image, and two images obtained from laterally resolved secondary ion mass spectrometry experiments are shown.
Anisotropic microstructures in ceramic green bodies result in nonuniform dimensional changes during sintering. Quantitative characterization of particle orientation together with processing/orientation relationships in these ceramic components are essential to control the dimensional variations during sintering. In this report, we discuss some practical aspects of measuring edge orientation in microstructures obtained by Field Emission Scanning Electron Microcscopy. Edges in the scanning direction were found to be less sensitive to the edge detection operator. Various edge sharpening algorithms were used to improve the edge contrast and hence achieve unbiased edge detection. Edges were detected using the Laplacian of Gaussian operator. Edge orientation polargrams were obtained by computing the intensities of detected edges with a 5×5 Prewitts operator convolved with a Gaussian. The computed edge orientation polargrams from the edge sharpened images were found to be internally consistent. Quantitative measure of particle orientation in tape cast ceramic microstructures was used for the first time to explain nonuniform sintering shrinkages.
A novel thinning algorithm for three-dimensional (3D) binary images is presented, with applications in the study of neuronal micro-anatomy by light microscopy. This algorithm satisfies properties important to many biological applications, including (a) connectivity preservation, (b) thinness, and (c) geometry preservation. It is fast to execute (a few minutes for typical 3D data sizes) on a personal computer. The algorithm addresses many challenges that are presented by 3D data. Algorithm improvements, over precursory algorithms, include the following: (1) Stricter and more exhaustive constraints on identifying outmost-layer border points are applied. (2) Border points are deleted by a novel ascending order of weighted neighbor count approach. The algorithm is robust in that it retains the above three properties (a–c) in the presence of relatively severe noise, uneven dye uptake, and nonuniform background.
We present a method of quantitative measurement, by laser scanning microscope, of intensity of fluorescence in assays on concentration of fluorochromes in a sample. This method can be used to evaluate the DNA content or the concentration of any protein, marked by an appropriate fluorescent dye, in single cells or groups of cells, isolated or in a tissue. Confocal laser scanning microscopy provides optical serial sections through thick biological sample that give accurate information about the intensity of fluorescence. The method we propose consists in scanning the sample by juxtaposition of optical sections and displaying it as a suitable projection. Multiplying the fluorescence intensity in such projection by the number of sections we obtain an estimate of the concentration of fluorochrome in the sample. We tested this method on fluorescent beads and on biological samples stained for DNA. We evaluated and corrected the measure for the photobleaching effect and for the beam theoretical distortion. Our results show that this method is accurate and suitable to compare samples of variable thickness and for multiple stained samples, and is less time consuming than the method based on the scanning of the sample according to the Nyquist principle.
A procedure for volume estimation based on scanning force microscopy images is applied to the study of banding-induced structural changes of chromosomes. Therefore, metaphase chromosomes were imaged before and after trypsin digestion, and the resulting three-dimensional data sets were used for a determination of the volumes of the imaged structures. The procedure is based on a histogram-based thresholding. The estimated volume is corrected for the background signal using the average background value from the histogram, so that an automated analysis of the images is possible. A first set of experimental data processed according to this approach is presented.
Vickers-type hardness patterns were investigated by atomic force microscopy. The resulting bitmap images were analyzed to evaluate the elastic relaxation of the impression. A procedure, which is suitable for the comparison of the shape of the residual hardness impression and that of the Vickers pyramid, is presented. With the help of difference images can be verified that the strongest relaxation takes place at the tip and at the horizontal edges of the Vickers pyramid.
Imaging secondary ion mass spectrometry (SIMS) is a powerful surface analysis tool capable of producing two-(2D) and three-dimensional (3D) spatially resolved images of element distributions. Both the 2D and the 3D imaging mode produce multispectral images, i.e., each image stack visualises the lateral distribution of one element, which divides the sample information into a number of individual images. Visual perception of the sample information is aggravated by this—for example, the exact spatial location of different elements relative to each other, formation of occlusions or segregations, etc.—is often hard to recognize when looking at n separate images, especially regarding 3D depth profiles. Image fusion is a process whereby images obtained from various sensors, or at different moments of time, or under different conditions, are combined together to provide a more complete picture of the object under investigation. The use of colour introduces a new dimension of information and can be used to simplify image analysis and object identification. This work introduces a complete methodology that enables the calculation and rendering of 3D colour images of multispectral SIMS depth-profiles based on the use of adaptive color and transparency maps. Examples of 3D SIMS images are given but the technique may be expanded to fusion and visualization of any other multispectral 3D image stack.
The public domain image analysis program NIH Image (http://rsb.info.nih.gov/nih-image) has been modified and extended to produce Image SXM (http://reg.ssci.liv.ac.uk), a program that supports many of the image file formats used by various scanning probe microscope manufacturers. This article discusses the different approaches to software customization and describes the use of two sets of macros in the analysis and processing of atomic force microscopy images.
The topic of this work is the study of direct detection of electrons by Charge-Coupled-Devices (CCD). The aim is to design a detector for the angle and energy-selective detection of signal electrons in very low energy scanning electron microscopy (VLESEM), using an electron-bombarded CCD sensor (EBCCD). We concentrate upon two problems—the design of appropriate electronics and determination of an appropriate energy of the signal electrons for the CCD sensor.
A simple algorithm is presented to obtain the area and the centroid of a plane homogeneous feature, by using only the coordinates of the boundary points. The algorithm also allow higher order moments to be calculated. From such centroid, and the centroid from edge vectors, we defined a shape descriptor invariant to Rotation, Scale, and Translation transformations. This parameter is able to characterize features with asymmetric distribution of roughness or reentrants.
The relative spatial distribution of proteins was investigated with immunofluorescent methods by confocal laser scanning microscopy and digital image restoration. For confocal data sets recorded with a voxel dimension of 50 × 50 × 150 nm noise and blur can be decreased and the resolution in the z -axis increased by applying the maximum likelihood estimation algorithm of the Huygens software. This approach was successfully applied to the study of tight and adherens junctions in relation to the actin cytoskeleton in Madin Darby Canine Kidney cells. Colocalization analysis was done for pairs of probes using a histogram-based method. F-actin, occludin, zonula occludens 1, and E-cadherin were included in the study. Double-labeled preparations were used. The combination of deconvolution with the colocalization of confocal data sets offers a powerful tool to investigate the spatial arrangement of proteins.
A new method is presented for the measurement of perimeter (length of boundaries) in binarized digital images. It considers the boundary to be represented by local configurations of pixels, smooths the pixel values, and constructs a super resolution contour line through them to measure the length of the perimeter. The results are more accurate and less sensitive to feature position and orientation than traditional perimeter measurements.
Accurate thresholding and segmentation of three-dimensional structures within thick biological specimens is particularly difficult to achieve. However, there exists a vast array of possible methods and approaches with which to tackle this problem. In this paper, we describe the problems associated with cellular segmentation and computerised analysis of confocal derived data from living blood vessels. In addition, current segmentation methods are examined and discussed in relation to their use with biological samples. Finally, a novel iterative multilevel thresholding and splitting method for semiautomated 3D segmentation of objects of different brightness and intensity homogeneity is presented. The method is particularly suited to segmentation of vascular cell volumes. The method has been tested on three typical confocal data sets, each of which have features common to 3D volumes of living biological tissue. The segmenter has been estimated to produce results which are accurate to within 90% of the “ground truth” measurements.
We have done a three-dimensional visualization of transgenic tobacco (Nicotiana tabacum) leaves for the study of chloroplast gene expression and regulation. The aim was to visualize tobacco leaves shot with tungsten particles. These tungsten particles were coated with the foreign DNA and shot into the leaf using the biolistic technique of DNA insertion. The visualization can be used to examine the leaves to gauge the efficiency of the shooting process, i.e.,to see what parts of the leaves have been effectively penetrated by the DNA-coated tungsten particles and also to judge the depth of penetration. The image data for the 3D visualization was collected at planes 10 microns apart, using a prototype version of a High Numerical Aperture Reflecting Microscope. The raw image data collected from the microscope was restored using the Row Action Projection (RAP) algorithm and the Partial Minimization and Constrained Iteration (PCMI) algorithm. These restored images were then used for 3D visualization using the Visualization Toolkit.
A Phillips webcam (price ∼ 70 USD) was mounted on an Olympus BX60 light microscope in order to record the images produced by the lens optical system. The quality of the images is comparable to those obtained by a standard one-chip charge coupled device (CCD) camera. The handling and implementation of the whole system has proven to be very easy due to the use of a normal micro computer and standard windows compatible software. Moreover, the presented method allows to record movies of investigated samples.
