Biotechnological applications of image analysis: present and future prospects

Biotechnological applications of image analysis: present and future prospects

Journal of Biotechnology, 23 (1992) 1-18 1 © 1992 Elsevier Science Publishers B.V. All rights reserved 0168-1656/92/$05.00 BIOTEC 00711 Review Bi...

1MB Sizes 0 Downloads 70 Views

Journal of Biotechnology, 23 (1992) 1-18

1

© 1992 Elsevier Science Publishers B.V. All rights reserved 0168-1656/92/$05.00

BIOTEC 00711

Review

Biotechnological applications of image analysis: present and future prospects Susan E. Vecht-Lifshitz and A n d r e w P. Ison SERC Centre for Biochemical Engineering, Department of Chemical and Biochemical Engineering, University College London, London, U.K. (Received 3 June 1991; revision accepted 28 August 1991)

Summary The current and potential biotechnological applications of image analysis and image processing systems are reviewed. Image analysis systems have proven to be highly versatile and efficient tools for assisting academic biotechnological research. It is expected that image analysis systems will allow more rapid and accurate quantification of numerous biotechnological analyses. There is, therefore, much scope for the implementation of image analysis/processing systems in a large variety of industrial and clinical applications. Image processing; Bioprocess monitoring; Cell morphology and motility; Histopathology; Enzymology; Cancer detection

Introduction Image analysis is an indispensable tool for microscopists who need to obtain accurate quantitative information from their samples (Joyce-Loebl, 1985). The advent of new video technology and increasingly powerful computers allows application of this technique to a very large range of methodologies. The effectiveness:cost ratio of computers is continuously rising and leads to the concomitant expansion of image analysis applications. Correspondence to: S. Vecht-Lifshitz, SERC Centre for Biochemical Engineering, Dept. of Chemical and Biochemical Engineering, University College London, Torrington Place, London WC1E 7JE, U.K.

Image analysis (IA) itself involves the digitization of the image into a grid of points or picture elements, pixels, and the measurement of the intensity of light at each point. Black and white images may be recorded by a single measurement of intensity (grey level) and colour images by measurement of the intensity at three wavelengths (red, green and blue; Jackman, 1989). A typical image may be digitized into 512 x 512 pixels with an intensity scale of values 0-255. Image analysis systems can detect different levels of light and intensity which the human eye cannot distinguish. There are several stages involved in grey level image analysis: image analysis, image acquisition, feature detection and measurement and data processing, and image storage. Image pixel values can be subjected to any mathematical operations required. This demands much computer memory and time. Typical image processing operations performed on a given image include shading correction, contrast enhancement, convolution, edge finding, fourier analysis and applying shape factors (reviewed by Jackman, 1989). Computer analysis systems have advantages over the human brain in terms of memory, quantitative measurement and repeating tasks. They are inferior in terms of interpretation and response to unexpected phenomena. There are at least 60 different U.S. companies which manufacture image processing and analysis systems. The evergrowing number of hardware (Bright, 1987; Digesu, 1989; Duller et al., 1989; Matsuyama, 1989; Sasov 1989; Pover, 1990) and software (Landy, 1988; Preston, 1988; Knight, 1989; Lennard, 1990; Morris, 1990) systems have been reviewed elsewhere. It has been suggested that the availability of user-friendly scientific image analysis software for the Macintosh ii has made the application of digital imaging techniques both practical and cost effective (Lennard, 1990). There is much potential to speed up the mathematical manipulation of image processing data by applying the latest computer technologies such as RISC processors and Acorn computers.

Current applications of image analysis The main areas of application of image analysis involving microscopy were listed by Pover (1990). These ranged from DNA sequencing, to asbestos monitoring, cell counting and gunshot residue analysis. Various biotechnological aspects of image processing and image analysis have recently been reviewed (Table 1). Of course, not every image analysis system is capable of all these applications, and the constraints imposed by hardware and software packages should be defined prior to purchase. Image analysis systems can also be used on the macroscopic scale in analysing geographical and astronomical data. IA is of particular use in analysing archeological and historical landviews. Today, there is a general consensus that there should be compatibility between the software packages in different image analysis systems, though in reality little has been done in this direction. It was recently stated that the lack of standardization of the software involved meant that the wheel was continually being reinvented (Preston, 1988). However, Landy (1988) suggested that Unix systems

TABLE 1 Recent reviews on image analysis and image processing First author

Title

Year

Mantas J.

Methodologies in pattern recognition and IA: a brief survey Multispectral magnetic resonance IA New trends of IA in the medical field Microscopy and IA for R and D (of food) An introduction to image processing in medical microscopy IA of single macromolecules Overview of flow-cytometryand IA in biological oceanography and limnology IA in quantitative cyto-pathologyand histopathology A review on biomedical image-processingand future trends IA techniques for automatic evaluation of 2-D electrophoresis

1987

Vannier M . W . Dengler J. Dziezak J.D. Douglas M.A. Frank J, Legendre L. Wied G.L. Dhawan A.P. Hader D.P.

1987 1988 1988 1989 1989 1989 1989 1990 1990

together with C language programming were becoming prevalent in the field and hence, a truly portable image processing package was becoming available. In this review, the present and future biotechnological applications of image analysis systems will be presented. This includes an overview of the current literature on image analysis in fields ranging from molecular biology to chemical engineering.

Areas of application in microbiology

Quantification of growth and cellular viability Using image analysis, colony counting on agar plates was performed more rapidly and accurately than with most manual processes. It was used to quantify growth rates of solid state cultures (Caldwell and Germida, 1985). The total white area (cell movement) was subtracted from the total black area (cell movement plus growth) to define the increase in area due only to growth. I A has also been applied to the enumeration and sizing of aquatic bacteria (David and Paul, 1989). Computer enhanced darkfield microscopy has been applied to the quantitative analysis of bacterial growth and behaviour on surfaces (Lawrence et al., 1989). Darkfield microscopy provided the potential advantages of a large observation field, large sample size and increased depth of field and ease of automation. These systems quantified colonization kinetics, bacterial growth rates, enumeration of cells, determination of rates of motility, number of motility events and determination of the recolonization kinetics of bacteria growing in surface microenvironments (Lawrence et al., 1989).

Image processing systems have been used to count plant-cell protoplasts and to determine their viability using FDA fluor staining, as well as measuring the concentration of their pigments (Zyrd et al., 1987). This technique was used to differentiate between viable and non-viable hybridoma cells (Frame and Wu, 1990). Using darkfield IA in a marine community, ciliates could be distinguished from bacterial prey on the basis of size and movement (Lawrence et al., 1989). Similarly, the enumeration of pathenogenic bacteria, and quantification of chlorine-induced bacterial injury were easily monitored using an image analysis system. It was shown that the IA method of determination of stressed E. coli cells was far superior to, and faster than that of the plate count method (Singh et al., 1990). The growth rates of different individual colonies in agarose has recently been determined using high resolution image analysis (Slocum et al., 1990). Morphological measurements of filamentous microorganisms have been determined using IA systems (Adams and Thomas, 1988; Packer and Thomas, 1990). Such systems quantified hyphal lengths, number of hyphal tips and hyphal length distributions. Biomass determination

Bjornsen (1986) reported that bacterioplanktonic biomass can be quantified using a simple factor based on the cell volume : biomass ratio. Similarly, Frame and Wu (1990) used cell volume measurement as an estimation of mammalian cell biomass employing an image analysis system. They suggested, however, that cell volume data should be used in conjunction with cell concentration data. DeYoung (1988) used IA for calculating cell areas and perimeters, as well as quantifying intercellular distances. Errors in measuring biomass using darkfield microscopy (Lawrence et al., 1989) were introduced when the developing biofilm became three dimensional. This made the measurement of area an inadequate estimate of biomass. It should be noted that illumination intensity of a microscope may introduce experimental errors by causing growth inhibition, morphological alteration, and death in bacteria. Lawrence et al. (1989) reported that darkfield microscopy could be used to determine biomass provided that bacterial growth was two dimensional. They compared the accuracy of various different systems, and pointed out that errors were introduced due to object blooming resulting from saturation of the camera photoreceptor, 3-D growth, and light scattering occurring at the edge of microbial cells (not necessarily proportional to the biomass increase). Packer and Lilly (1991) showed that an IA system could be used to measure the area of mycelial organisms. The volume of these microorganisms was assumed to be = ~- x hyphal diameter × cross section area x 0.25. The biomass was calculated by multiplying the density of the various cellular components by their fractional volume. This method was accurate to + / - 8% up to a biomass concentration of 38 g 1-1. It has been reported that IA was a practical technique for indirect measurement of the biomass concentration present in a culture (Frame and Wu, 1990). Predicted values of mammalian cell biomass from volume determination

were accurate to 11%. The mean volume of the viable cells appeared to vary with their growth rate (faster growing cells being larger). Determination of plant biomass by IA has also been reported (Evers et al., 1987). Hiraoka (1986) published a patent on the determination of fungal biomass contained in sewage. However, it should not be forgotten that there are cheaper and easier methods of continuously determining non-filamentous biomass (Cox et al., 1989) other than image analysis, such as turbidimetric methods. Recently, a patent was issued for biomass determination using electrodes to define the voltage:current ratio (Kell and Todd, 1989). The long term potential of IA systems would be to quantify real-time biomass concentrations during bioprocesses, including filamentous, pelleted and viscous systems. This would require development of an on-line sampling system and employing faster methods of capturing and storing data. The updated technology would probably require some capital investment in extra software and hardware. Such systems would be used to simultaneously log numerous growth and morphological parameters (possibly the uptake of solid substrates) and would therefore be of more use than other biomass determination techniques.

Cell sensitivity and antibiotic screening IA has been applied to antibiotic sensitivity testing using a control plate and test plates containing various concentrations of the antibiotic. This can be used to determine minimum inhibitory concentrations (MICs) and for screening urines for bacterial growth. IA can quantify microcolony inhibition and can be used to determine the sensitivity of microorganisms within 4 h. This is much faster than most other techniques available today (Hammonds and Adenwala, 1990).

Microbial adhesion, morphology and motility IA has been applied to quantify biofilm formation and early colonization of smooth substratum (Escher and Characklis, 1988). The technique employed allowed direct measurement of independent processes contributing to the colonization of the substrate. The behavioral characteristics of the microorganisms at the substratum e.g. growth rate and direction of motion orientation can be measured in situ. This technique further allows novel quantitative analysis of spatial distributions of organisms during adsorption. It was found in this study that sorption related processes depend on shear stress and bulk CFU concentration, whereas growth related processes depend on cell concentration at the substratum alone. It was concluded that sorption related processes are zero order with respect to the substratum whereas, growth related processes were first order reactions with respect to the substratum concentration. The measurement of rapid in vitro attachment of Candida albicans cells to transparent acrylic has been monitored using a Magiscan IA system (Shakespeare and Verran, 1988). IA can readily quantify the attachment of yeast cells, providing a more accurate and fast methodology than manual counting. This technique can

be applied to measuring attachment to medically important surfaces, for example: intravenous devices, catheters and prostheses. Sjollema et al. (1989) developed a parallel plate flow cell to perform real-time enumeration of adhering microorganisms. A novel application of image processing was to observe chemotaxis in Dictyostelium discoideum (Fisher et al., 1989). It was found that the amoebae could spatially integrate information about local cAMP levels at various points of their cell surface. The dependence of the chemotactic response and motility as a function of the cAMP concentration and gradient steepness were quantified. A novel chemotaxis chamber was described, which could also be applied to investigation of chemotaxis of leucocytes and other cells.

Modelling of mycelial morphology and metabolism Filamentous fungi and bacteria are microorganisms which can grow in dispersed viscous submerged cultures, or can aggregate to form pellets of varying sizes and compactness (Metz, 1976; Vecht-Lifshitz et al., 1990). The tendency to aggregate depends on genetic as well as environmental factors and can be controlled by manipulating the growth conditions (Braun and Vecht-Lifshitz, 1991). An understanding of the microbial physiology is required so that industrial processes can be modelled, controlled and optimised. Image analysis and microscopic systems could be used to determine the location of production and excretion of coloured antibiotics. This would give some insight into the optimal morphology for a specific antibiotic. Conversely, the influence of growth conditions and medium composition on productivity and morphology could be quantified using standard colorimetric and novel IA techniques. A few recent fundamental studies have investigated the developmental morphology of filamentous microorganisms using image analysis. Reichl et al. (1990a) measured the interseptal spacing in hyphae in Streptomyces tendae. Accurate positional information about septa and branches was obtained. It was found that both the mean length and the branch rates changed significantly with time. Normally only 1-3 branches were observed between septa in the first 10 h of cultivationl Morrin and Ward (1989) used a fluorescent marker (Calcofluor White) to visualize extension zones and branch points in Rhizopus arrhizus as a function of the culture pH and presence of polymer. Apical hyphal extension can be quantified using immunofluorescent labelling of wall components, or using tritiated N-acetyl-D-glucosamine (Gray et al., 1990). Light microscopic autoradiography showed that tritiated GlcNAc was preferentially incorporated at hyphal tips and to a lesser extent, at apices of lateral branches. It also showed that there was a small degree of wall synthesis, turnover or thickening. A mean extension rate of 8.03 microns h-~ was noted for Streptomyces coelicolor in submerged culture, and was similar to the extension rate in solid medium of 10-13 microns h -1 (Gray et al., 1990).

The average extension rate in a specific bioprocess should be defined using image analysis. The influence of this extension rate upon the productivity should be determined. The effect of the branch rate upon productivity could be investigated by manipulation of the growth rate by changing the medium composition. In addition, there are drugs such as Hinosan which reduce hyphal extension and increase the branch rate without affecting the growth rate (Wiebe et al., 1990). This could be added to the growth at various times to see whether productivity was morphology dependent or independent. Kotov and Reshetnikov (1990) described a stochastic model which applied to exponential mycelial growth. Three different types of hyphal cells were considered: apical, branching and growth cessation. Theoretical calculations compared well to actual growth patterns in Streptomyces coelicolor and Rhizoctonia solani. Obert et al. (1990) described microbial growth patterns by fractal geometry employing an image processor linked up to a light microscope. They showed that fractal geometric patterns could be applied to fungal and bacterial growth. They further showed that multibranched mycelia were similar. This implied that the global structure of an object may be complex, although the fundamental growth concept may be very simple. The models should be developed to predict complex mycelial structures (pellets) as a function of time and growth conditions. Considering that mechanisms for growth were similar in both solid and liquid media in Streptomyces tendae (Reichl et al., 1990b), it may be that the mechanism of metabolite excretion is the same in liquid and solid cultures. One would hope that such studies would lead to optimisation of production under given conditions and optimisation of conditions to yield more product. An appropriately devised sampling module for on-line mycelium biomass withdrawal would allow growth rate measurement and morphology characterization. If the growth rate were too high, the dissolved oxygen concentration could be reduced, feeding could be reduced, or the temperature could be reduced, and vice versa. If the cells were aggregating too much, the shear rate could be increased, and the pH could be changed, and surfactants or polymers could be added (Vecht-Lifshitz et al., 1989). Depending upon the mechanism of cellular aggregation, other variables could be changed. A secondary spore/vegetative inoculum could be added during the growth. If solids would aid the aggregation process, they could be added during cultivation. At low productivities, it could be scrapped earlier rather than later, and the losses would be minimised. The biomass concentration in a bioreactor is normally less important than the metabolite concentration (apart from when it is the product). It should therefore be considered as a manipulable parameter like the DO, and not as an end result itself. A recent review on secondary metabolism regulation in .4ctinomycetes suggested that secondary metabolism might be a physiological expression of cell differentiation (Horinouchi and Beppu, 1990). The various diffusible low-molecular weight chemical substances which act like hormones for secondary metabolism and differentiation were described. Granozzi et al. (1990) showed that there is physiological stress within a bacterial population before the onset of morphological differentiation. Growth almost ceased before the synthesis of aerial mycelia,

whereas mutants do not show this phenomenon. The stress is not induced by depletion of the nutrient within the medium. This phenomenon is similar to the stringent response mechanism in antibiotic production in Streptomyces (Ochi, 1986). The study of streptomycetal stringent response and its correlation with the major metabolic processes might be useful in yielding an understanding of the molecular mechanisms that trigger differentiation in these bacteria. One might predict from this evidence that the genes for aerial mycelial differentiation are on the same gene cluster as those for the secondary metabolites, and are probably activated by the same factors (Horinouchi and Beppu, 1990) at the same time. If this were the case, one might try to quantify the appearance of aerial mycelia using IA on solid mycelial cultures to predict the quantity of secondary metabolite to be produced. Some strains might lose the ability to sporulate without losing the ability to produce a secondary metabolite, so care is required in such studies.

Applications of IA in plant cells and plant cell cultures IA is an extremely versatile tool for analytical diagnosis of stress induced reactions in plants (Omassa et al., 1987). Fluorescent chlorophyll transients can be used for diagnosing the photosynthetic systems of attached leaves. The obvious advantage of such systems is the ability to investigate what happens to the plant in situ. IA can be applied to the non-invasive evaluation of growth during plant micropropagation (Smith et al., 1989). The degradation of plant tissue has also been quantified using image analysis (Twidwell et al., 1989).

Applications of image analysis in biomedical research The progress of biomedical research over the past decade has been enormous. Practical systems for medical imaging were developed over the last 10-15 years such as those by Elscint, Haifa, Israel. Ram described a system for image processing in nuclear medicine as early as 1979. The combination of novel techniques in molecular biology and immunology coupled with advances in microprocessor and computer technology have led to many breakthroughs in the field of enzymology, histopathology, molecular biology, as well as cancer genetics, mutagenesis and oncology (Tanke, 1989).

Enzymology Julis et al. (1987) demonstrated the potential usefulness of IA in enzyme histochemistry. He suggested that more parameters, useful for the evaluation of enzymic reactions can be obtained than those from densitometric methods. IA allows quantitative capture of the heterogeneity of enzyme equipment of brain capillaries. It can also be used to determine DNA content and various enzyme activities. Van Norden (1990) reviewed the application of image analysis to in situ measurement of enzyme reactions. Enzymes occur in the cytoplasmic matrix

TABLE 2 Determination of protein/enzyme structure and function by image analysis Enzyme

Investigation

Reference

Alcohol oxidase Alk. phosphatase ATP synthase Cholinesterase Dynein ATPase E. coli reca protein F1-ATPase Hemoglobin Hexokinase Immunoperoxidase Na, K-ATPase Proteinase Wheat-protein

EM and IA Structure by IA Structure by EM and IP Activity by IA Stereo images by EM Structural domains by IA Structure by EM and IP Oxygen saturation Activity by IA Stain quantitation Digital IA EM and IA Electrophoresis

Vonck and Vanbruggen, 1990 Kaplow et al., 1986 Tsuprun et al., 1989 Dubovy et al., 1990 Burgess et al., 1990 Yu and Egelman, 1990 Boekema et al., 1986 Hashimoto et al., 1987 Lawrence et al., 1989 Pullman and Bur, 1989 Beall et al., 1989 Baumeister et al., 1988 Dougherty et al., 1988

EM, electron microscopy;IP, image processing; IA, image analysis.

(which behaves as a non-ideal solution). Association of the enzyme with other enzymes or with the cytoskeleton can cause changes in enzymic kinetic p a r a m e t e r s and can change the reaction velocity by several orders of magnitude. Fluorometric m e a s u r e m e n t of enzyme reactions in individual cells have several advantages over chromogenic reactions. The sensitivity and spatial resolution in such systems are high and distributional errors do not occur. These techniques have led to the recent development of analysis of substrate concentrations in situ. Furthermore, cytophotometric data of metabolic activity and enzyme reactions allow correlation of quantitative information with morphological data. Tools are now available to determine quantitatively the characteristics of enzymes and metabolic pathways in situ, in conjunction with the morphology and structure of tissues and ceils. Moreover, the methods can be used to analyze post-translational regulation mechanisms of cell metabolism and alterations due to pathological conditions. The great advantage of an image analyzer over scanning and integrating cytophotometers is the capacity to store 2-D images of enzyme reactions as they proceed with time in a microscopic field. Some of the latest studies on e n z y m e / p r o t e i n structure and function are displayed in Table 2.

Molecular biology and genetic engineering Image processing systems are useful in speeding up the screening c D N A libraries (Lu et al., 1990) and for picking up differences in signal intensities displayed by specific plaques. Such systems can spot rare events such as clones with in vitro transcripts, or clones with first strand c D N A (both about 1%). This sensitivity is of great advantage, as the amount of biological material is often limited or difficult to prepare (Lu et al., 1990).

10 Gene screening can be performed without need to express or purify the gene product. Using IA, one can find genes that have been repressed, induced and those that are globally regulated (Knight, 1989). Methods to study the human genome at the single gene level are leading to discoveries in clinical genetics and oncology and in pre-natal diagnosis. Automated detection of rare events, such as mutant cells in frequencies lower than 1 in a million have become possible using automated image analysis techniques. This has been applied to finding bladder cancer cells, and has been used to complement flow cytometry studies (Parry and Hemstreet, 1988). Analyses involving radioisotopes are gradually being replaced by fluorescent assays. Technical developments in laser technology, image processing systems, and charge coupled device detectors, have led to the possibility of detecting one fluorophore. Novel applications of fluorescent techniques using a light microscope and image processing systems can now detect and quantify DNA adducts (compounds that bind to DNA) at almost the single molecule level. Specific antibodies against these adducts have been raised and this has led to the localisation of the adduct in the cell nucleus. Further applications include detection of minute amounts of macromolecules (DNA or proteins) using cytochemical staining, in situ hybridization studies and rare event detection (Tanke, 1989). New Ca 2+ sensitive fluorescent indicators coupled to IA systems have allowed measurements of transients within single cells (Leong, 1989). Novel spatial and temporal elements of the cellular signalling apparatus can now be investigated with such systems. These findings will hopefully lead to breakthroughs in research of human cellular signalling mechanisms such as pituitary function.

Histopathology and neuropathology The potential of image analysis is well demonstrated in the field of histo/ cytopathology. The ultimate goal is predictive reliable diagnosis of rare events, such as cancer. Using data bases and IA studies, various functions in a patient's histopathology were quantified. A large group of patients was required for sampling. Various discriminant functions were quantified using micromorphometry and were plotted against each other (Wied et al., 1989). Q-analysis was utilised, allowing an examination of the relationship between two sets of data. Hence, regions of normal and abnormal cytology were determined. Such studies applied to intermediate cells, could lead to diagnostic clues in about 70% of patients with dysplastic or malignant disease. Clearly, these tools are invaluable for predicting disease, but today, the lag in transfer of research to clinical practice is a major bottleneck (Wied et al., 1989). Image processing systems have been used to investigate histological sections of nervous tissue (Schleicher and Zilles, 1990). The stereological methods reported were reliable and unbiased, and led to localisation of the boundaries of brain regions. It is thought that much effort is yet required to combine stereological methods with IA. It is hoped that this will lead to a more precise understanding of neural anatomy and function.

11

Human and mammalian cell technology IA has been applied to differentiate between viable and non-viable sperm (Hall, 1990). Perhaps this will ultimately lead to systems for the selection and removal of imperfect sperm, and hence improve in vitro fertilization techniques. Developments in this area might lead to the selection of healthy zygotes prior to implantation. This would be of immense importance to improving human and animal fertility. Epidermal growth factor (EGF) and platelet derived growth factor (PDGF) were added to a live mammalian cell culture. It was noted that the addition of both platelet derived growth factor and epidermal growth factor quantitatively influenced the cell motility and induced morphological changes in living cultured cells. This type of study can be used in the research of the factors which induce cancer and abnormal cell proliferation. Digital image processing allowed the quantification of the cell motility and morphological changes (Miyata et al., 1988). Ability to quantify such changes should lead to new research frontiers in determining what factors induce or switch-on the development of cancerous cells from normal ones. The sensitivity of different groups of cells to their environment, e.g. pH, can be monitored using digital imaging systems. In fact, such systems may, in the future, be of help in predicting which cells may turn into cancer ceils, by having nonstandard motility. It is hoped that this type of system may be used to test anti-cancer and other drugs on mammalian cells in vitro and this would speed up standard drug testing procedures and possibly replace using animals for drug testing. It is also more reliable to test drugs on human cells than in animal models, as the two do not necessarily respond in the same way. IA has been used to evaluate the in vivo oxygenation of red blood cells. This was performed by dual video systems and two wavelength analysis (Ellis et al., 1990). Another use of digital image processing is the observation of cellular responses to stimuli in situ. Paradiso et al. (1987) applied intracellular dyes to measure cytosolic free calcium and pH i in gastric oxyntic and chief cells. The two groups showed different responses to acid/alkaline pH change. It was noted that oxyntic ceils responded differently from chief ceils (Paradiso et al., 1987). Again, this kind of tool could be of use in a large variety of in situ experiments, where morphology and cell motility are of interest. Image analysis can be used to determine the shape of cells and cell components. Simmons and Richards (1988) measured various lung cell volumes and found that the nuclei were asymmetric spheroids (as oppposed to spheres). Image analysis can further be developed to take several 2-D slices of a system and then to reconstruct the third dimension. The theory and methodology of 3-D reconstruction are thoroughly described by Engelhardt (1988). Image restoration and reconstruction artefacts were considered. This type of image processing has been applied to 3-D karyotype analysis. Interestingly, it was concluded that there is a non-random localisation of the chromosomes with respect to each other (Jones et al., 1990). The shape and motility of live cells in culture or those preserved histologically can be viewed. Quantitative video microscopy can be applied to monitor cell motility

12 and hence one can discern between normal cells and those with metastatic potential (Knight, 1989). 3-D video technique can be coupled with cell-labelling using fluorine or bromine. This can be used for intracellular localization of drugs in cultured tumour cells (Berry et al., 1990).

Image systems for food, biomaterials and biotechniques analysis Food

There are numerous applications of IA in food technology. Areas investigated to date include microbial enumeration, raw materials analysis, milk and chocolate powder sizing, emulsion characterization and analysis of tea-bag perforations (Dziezak, 1988). There are many other areas to which this might be applied in the food scene, particularly in quality control, texture and fluid characterization and the like. Fernandes et al. (1987) used IA to determine threshold levels of microbial contaminants of solid food. Methods of automation of this process and better means of distinguishing between bacteria and other objects are being developed by his team. Biomaterials

Jokela et al. (1990) used computerised microscopic image analysis to determine emulsion droplet size distribution. They described the advantages of the image analysis system over other systems, such as the Coulter counter as (1) being able to determine the particle size of a large range of sizes, 1 micron up to macroscopic; (2) having quicker capture of samples; (3) no need to dilute samples; (4) non-conducting samples can be investigated; (5) one can distinguish between coalesced and flocculated droplets. The development of microstructural descriptions of heterogeneous materials was described by Saltzman et al. (1987). Zygourakis and Glass (1988) presented a study on structure of macropores in coal chars. They claimed that the structure of the macropores could not be defined by other indirect analytical methods. This type of analysis could be further applied to investigating the structure of ultrafilter and other membranes, membrane reactors and observation of how membranes become clogged up with continuous use. Vivier et al. (1989) reported the study of microporous membrane stucture employing an image analysis system. Similar applications could be used in examining biological phenomena of the skin (Barton and Marks, 1988), of fibres and food structures, immobilized catalysts, enzymes and cells. Hanzevack (1988) reported the use of a very cheap ($13,000) and versatile system for measuring 2-phase (liquid-liquid) flow. Digital image processing is used to measure drop dispersion and mixing in highly turbulent two-phase flow in a pipeline, employing pulsed laser signals.

13 TABLE 3 Analytical applications of image analysis Methodology DNA quantification DNA quantitation Flow and image cytometric DNA analysis Fluorescent biogenic amines quantification In situ hybridization quantification Microdensitometry Nonradioactive in situ hybridization Nucleic acid blot analyzer Quantitation of antigen in tissue by immunofluorescence image analysis Papanicolaou and Feulgen staining Quantitative karyotyping Restriction enzyme fingerprint autoradiograms Size and molecular weight by density gradient ultracentrifugation Thin-layer chromatographic plates TLC of polyamines Three-dimensional chromosome analysis Two-dimensional electrophoresis Two-dimensional electrophoresis gels Two-dimensional polyacrylamidegels

Reference Fausel et al., 1990 Bauer et al., 1989 Kowalvern et al., 1990 Sato and Shibuya, 1990 Moro et al., 1990 Sanchez et al., 1990 Larsson and Hougaard, 1990 Auron et al., 1988 Basgen et al., 1989 Gurley et al., 1990 Fukui and Kakeda, 1990 Sulston et al., 1989 Legoff et al., 1990 Jansen et al., 1989 Wettlaufer and Weinstein, 1988 Jones et al., 1990 Hader and Kauer, 1990 Lee et al., 1988 Moline and Hruschka, 1987

Biotechniques D e Y o u n g (1988) reviewed the use of I A systems for automatic analysis of electrophoresis gels, m e m b r a n e s and dry films. H e suggested that one should first choose all the software one requires, and then buy the hardware which can process the given software. This recommendation is in view of the fact that people tend to do the opposite and then often get stuck when they want to introduce some other software package. Image analysis can also be used for a large variety of analytical measurements (Knight, 1989). Its speed of image capture allows it to be applied to analysis of blots, wells, microscopic images and fluorescent gels. Belchamber et al. (1987) showed that image processing could be applied to quantification of separations of high performance thin layer chromatographic plates. The advantages of using image processing systems for analysis of the plates are that this speeds up the process significantly, data is acquired much faster and there are no moving parts of the apparatus. In contrast, the obvious disadvantages are the system price, the requirement of large computer systems and the poor response of the system to the U V region of the spectrum. Some of the many analytical applications of I A are presented in Table 3. On-line image analysis systems may play an important role in the future in quality control. The microscopic image of a biological (or other) product could be

14

compared with an image of a control sample. Applications could be in disposable biosensor technology, for checking microcircuitry; for examining components for biological assay kits; or for immobilized biocatalyst structures.

Conclusions Image analysis and processing systems will have a profound effect on biotechnological research in the future. These systems ought to be unified to allow easier interaction between research groups and practical comparison of reported studies. Image analysis will have significant impact on medical diagnoses in the 90's, on food quality control, determination of particulates, emulsions and fibrous membranes and on mycelial morphology and productivity interactions.

Acknowledgements The authors would like to thank Dr. M. Bushell of the School of Biological Sciences, University of Surrey, for reading the manuscript and for his helpful comments. The authors acknowledge the SERC for their support.

References Adams, H.L. and Thomas, C.R. (1988) The use of image-analysis for morphological measurements on filamentous microorganisms. Biotechnol. Bioeng. 32, 707-712. Auron, P.E., Galson, D.L., Fenton, M.J., Clark, B.D., Cole, E.S., Peters, L., Sullivan, D. and Teller, D. (1988) The nucleic-acid blot analyzer. 2. Analyze, an image-analysis software package for molecular-biology. Biotechniques 6, 347. Barton, S.P. and Marks, R. (1988) Image-analysis as a tool for measuring biological phenomena of the skin. Int. J. Cosmet. Sci. 10, 137-144. Basgen, J.M., Michael, A.F. and Nevins, T.E. (1989) Quantitation of antigen in tissue by immunofluorescence image-analysis. J. Immunol. Methods 124, 77-83. Bauer, T.W., Tubbs, R.R., Gephardt, G., Edinger, M. and Levin, H.S. (1989) A prospective comparison of DNA quantitation by image-analysis and flow-cytometry. Lab. Invest. 60, A7. Baumeister, W., Dahlmann, B., Kuehn, L., Kopp, F., Hegerl, R. and Pfeifer, G. (1988) Electron-microscopy and image-analysis of the multicatalytic proteinase. FEBS Lett. 241,239-245. Beall, H.C., Hastings, D.F. and Tingbeall, H.P. (1989) Digital image-analysis of two-dimensional Na,K-ATPase crystals - dissimilarity between pump units. J. Microsc. 154, 71-82. Belchamber, R.M., Read, H. and Roberts, J.D.M. (1987) Intracellular-localization of drugs in cultured tumor-cells by ion microscopy and image-processing. J. Chromatogr. 395, 47-53. Berry, J.P, Lespinats, G., Escaig, F., Boumati, P., Tlouzeau, S. and Cavallier, J.F. (1990) Intracellularlocalization of drugs in cultured tumor-cells by ion microscopy and image-processing. Histochem. 93, 397-400. Bjornsen, P.K. (1986) Automatic-determination of bacterioplankton biomass by image-analysis. Appl. Env. Microbiol. 51, 1199-1204. Boekema, E.J., Berden, J.A. and Vanheel, M.G. (1986) Structure of mitochondrial F1-ATPase studied by electron-microscopy and image-processing. Biochim. Biophys. Acta 851,353-360.

15 Braun, S. and Vecht-Lifshitz, S.E. (1991) Mycelial morphology and metabolite production. Trends Biotechnol. 9, 63-68. Bright, D.S. (1987) A lisp-based image-analysis system with applications to microscopy. J. Microsc. 148, 51-87. Burgess, S.A., Dover, S.D. and Woolley, D.M. (1990) Stereo images of cryofixed dynein ATPase and a method for their enhancement by digital image-processing. J. Physiol. 425, 6. Caldwell, D.E. and Germida, J.J. (1985) Evaluation of difference imagery for visualizing and quantifying microbial growth. Can. J. Microbiol. 31, 35-44. Cox, R.P., Miller, M., Nielsen, J.B., Nielsen, M. and Thomsen, J.K. (1989) Continuous turbidimetric measurements of microbial cell density in bioreactors using light emitting diode and a photodiode. J. Microbiol. Methods 10, 25-31. David, A.W. and Paul, J.H. (1989) Enumeration and sizing of aquatic bacteria by use of a silicon-intensified target camera linked image-analysis system. J. Microbiol. Methods 9, 257-266. Dengler, J., Desaga, J.F., Bertsch, H. and Schmidt, M. (1988) New trends of image-analysis in the medical field. Methods Info. Med. 27, 53-57. DeYoung, H.G. (1988) Microscopy and image analysis. Bio/Technology 6, 78-79. Dhawan, A.P. (1990) A review on biomedical image-processing and future-trends. Comp. Methods Programs Biomed. 31, 141-183. Digesu, V. (1989) An overview of pyramid machines for image-processing. Inform. Sci. 47, 17-34. Douglas, M.A. and Trus, B.L. (1989) An introduction to image-processing in medical microscopy. Med. Prog. Technol. 15, 109-140. Dougherty, D.A., Mattern, P.J., Zeece, M.G. and Wehling, R.L. (1988) Evaluation of wheat-protein quality by 2 dimensional electrophoresis with quantitative image-analysis. Cereal Food World 33, 665. Dubovy, P., Svizenska, I. and Vega, J.A. (1990) Nonspecific cholinesterase activity in mouse spinal ganglia - the usefulness of histochemical-study and image-analysis for simple characterization of neuron subclasses. Cell. Mol. Biol. 36, 23-40. Duller, A.W.G., Dagless, E.L., Thomson, A.R. and Storer, R.H. (1989) An associative processor array for image-processing. Image Vis. Comput. 7, 151-158. Dziezak, J.D. (1988) Microscopy and image-analysis for R-and-D. Food Technol. 42, 110-124. Ellis, C.G., Ellsworth, M.L. and Pittman, R.N. (1990) Determination of red-blood-cell oxygenation in vivo by dual video densitometric image-analysis. Am. J. Physiol. 258, 1216-1223. Engelhardt, H. (1988) Correlation averaging and 3-D reconstruction of 2-D crystalline membranes and macromolecules. In: Norris and Ribbons (Eds.), Methods in Microbiology, Vol. 20, Academic Press, London. pp. 357-413. Escher, A.R. and Characklis, W.G. (1988) Microbial colonization of a smooth substratum - a kinetic-analysis using image-analysis. Water Sci. Technol. 20, 277-283. Evers, H., Ernst, D., Liebig, H.P., Patzold, G. and Seibold, H.W. (1987) Determination of plant biomass by image-analysis. Gartenbauwissenschaft 52, 119-124. Fausel, R.E., Burleigh, W. and Kaminsky, D.B. (1990) DNA quantification in colorectal-carcinoma using flow and image-analysis cytometry. Anal. Quantat. Cytol. Histol. 12, 21-27. Fernandes, M.A., Clark, S.A. and Jackman, P.J.H. (1987) Detection of bacteria in solid foods by image-analysis. J. Appl. Bacteriol. 63, R2-R3. Fisher, P.R., Gerisch, G. and Merkl, R. (1989) Quantitative-analysis of cell motility and chemotaxis in Dictyostelium discoideurn by using an image-processing system and a novel chemotaxis chamber providing stationary chemical gradients. J. Cell Biol. 108, 973-984. Frame, K.K. and Wu, W.S. (1990) Cell volume measured as an estimation of mammalian cell biomass. Biotechnol. Bioeng. 36, 191-197. Frank, J. (1989) Image-analysis of single macromolecules. Electron Microsc. Rev. 2, 53-74. Fukui, K. and Kakeda, K. (1990) Quantitative karyotyping of barley chromosomes by image-analysis methods. Genome 33, 450-458. Granozzi, C., Billetta, R., Passantino, R., Sollazzo, M. and Puglia, A.M. (1990) A breakdown in macromolecular synthesis preceding differentiation in Streptomyces coelicolor A3(2). J. Gen. Microbiol. 136, 713-716.

16 Gray, D.I., Gooday, G.W. and Prosser, J.I. (1990) Apical hyphal extension in Streptomyces coelicolor A3(2). J. Gen. Microbiol. 136, 1077-1084. Gurley, A.M., Hidvegi, D.F., Bacus, J.W. and Bacus, S.S. (1990) Comparison of the Papanicolaou and Feulgen staining methods for DNA quantification by image-analysis. Cytometry 11,468-474. Hader, D.P. and Kauer, G. (1990) Image-analysis techniques for automatic evaluation of 2-dimensional electrophoresis. Electrophoresis 11,407-415. Hall, N. (1990) Image-analysis picks up the able semen. N. Sci. 125, 34. Hammonds, S.J. and Adenwala, F. (1990) Antibiotic-sensitivity testing of bacteria by microcolony inhibition and image-analysis. Lett. Appl. Microbiol. 10, 27-29. Hanzevack, E.L, (1986) Concentration by laser image-processing. Chem. Eng. Prog. 82, 47-50. Hashimoto, M., Fukuda, M., Hata, R., lsomoto, A. and Tyuma, I. (1987) Color analysis method for estimating the oxygen-saturation of hemoglobin using an image-input and processing system. Anal. Biochem. 162, 178-184. Hiraoka, A.M., Kazushi, T., Baba, K., Nogita, S. and Mori, S. (1986) Determination of fungal biomass contained in sewage. U.S. Pat. 456444 Jan. 14th 1986. Horinouchi, S. and Beppu, T. (1990) Autoregulatory factors of secondary metabolism and morphogenesis in Actinomycetes. CRC Crit. Rev. Biotechnol. 10, 191-204. Jackman, P.J.H, (1989) Image analysis. In: Bryant, T.N. and Wimpenny, J.W.T. (Eds.), Computers in Microbiology, IRL Press, Oxford. Jansen, E.H.J.M., Stephany, R.W., Vanlook, L.J., Vanpeteghem, C. and Vandenbosch, D. (1989) Quantitafive-analysis of anabolics on thin-layer chromatographic plates using image-analysis techniques. J. Chromatogr. Biomed. Appl. 489, 205-212. Jokela, P., Fletcher, P.D.I., Aveyard, R. and Lu, J.R. (1990) The use of computerized microscopic image-analysis to determine emulsion droplet size distributions. J. Coll. Inter. Sci. 134, 417-426. Jones, S.J., Taylor, M.L., Baarslag, M.W., Krol, J.J., Mosterd, B., Mans, A., Brakenhoff, G.J. and Nanninga, N. (1990) Image-processing techniques for 3-D chromosome analysis. J. Microsc. 158, 235 -248. Joyce-Loebl (1985) In: Image Analysis: Principles and Practice, Newcastle U.K. pp. 3-22. Julis, 1., Lodja, Z. and Schnabel, R. (1987) Some examples of the usefulness of image-analysis in enzyme-histochemistry. Histochem. J. 19, 600. Kaplow, L.S., Crouch, J.Y., Garcia, G.L., Kunz, H. and Meyers, J.A. (1986) Assessment of leukocyte alkaline-phosphatase by image-analysis. Ann. N.Y. Acad. Sci. 468, 85-92. Kell, D.B. and Todd, R.W. (1989) Determination of Biomass. US Patent 4810650 March 7th, 1989. Knight, P. (1989) Innovations in image-analysis. Bio/Technology 7, 954-956. Kotov, V. and Reshetnikov, S.V. (1990) A stochastic-model for early mycelial growth. Mycol. Res. 94, 577-586. Kowalvern, A., Gonzalezcrussi, F., Turner, J., Trujillo, Y.P., Chou, P., Herman, C., Castelli, M. and Walloch, J. (1990) Flow and image cytometric DNA analysis in rhabdomyosarcoma. Cancer Res. 50, 6023-6027. Landy, M (1988) Image-processing packages. Nature 335, 19. Larsson, L.I. and Hougaard, D.M. (1990) Optimization of nonradioactive in situ hybridization image-analysis of varying pretreatment, hybridization and probe labeling conditions. Histochem. 93, 347-354. Lawrence, G.M., Matthews, J.B. and Beesley, A.C.H. (1989a) The use of continuous monitoring and computer-assisted image-analysis for the histochemical quantification of hexokinase-activity. Histochem J. 21,557-564. Lawrence, J.R., Korber, D.R. and Caldwell, D.E. (1989b) Computer enhanced darkfield microscopy for the quantitative analysis of bacterial growth and behaviour on surfaces. J. Microbiol. Methods 10, 123-138. Lee, C., Hu, S.E., Lok, M.S., Chen, Y.C. and Tseng, C.C. (1988) Microcomputer-based image-analysis systems for two-dimensional electrophoresis gels. Biotechniques 6, 216-224. Legendre, L. and Yentsch, C.M. (1989) Overview of flow-cytometry and image-analysis in biological oceanography and limnology. Cytometry 10, 501-510. Legoff, D., Nouvelot, A. and Chermant, J.L. (1990) Determination of size and molecular-weight

17 distributions of lipoproteins using automatic image-analysis and density gradient ultracentrifugation. J. Biochem. Biophys. Methods 20, 247-258. Lennard, P.R. (1990) Image-analysis for all - software review. Nature 347, 103-104. Leong, D.A. (1989) Modern trends in image-analysis. Clin. Chem. 35, 1052-1053. Lu, X., Dengler, J., Rotbarth, K. and Werner, D. (1990) Differential screening of murine ascites cDNA libraries by means of in vitro transcripts of cell-cycle-phase-specific cDNA and digital imageprocessing. Gene 86, 185-192. Metz, B. (1976) From pulp to pellet. PhD Dissertation, Univ. Delft, Holland. Mantas, J. (1987) Methodologies in pattern-recognition and image-analysis - a brief survey. Pattern Recognit. 20, 1-6. Matsuyama, T. (1989) Expert systems for image-processing - knowledge-based composition of imageanalysis processes. Comp. Vis. Graph. Im. Proc. 48, 22-49. Miyata, Y, Sakai, H. and Nishida, E. (1988) Growth factor-induced and phorbol ester-induced changes in cell morphology analyzed by digital image-processing. Exp. Cell Res. 175, 286-297. Moline, H.E. and Hruschka, W.R. (1987) Computer-enhanced image-analysis of bacterial polypeptide patterns on two-dimensional polyacrylamide gels. Phytopathology 77, 745-747. Moro, L., Colombi, M., Crespi, S., Dilernia, R. and Barlati, S. (1990) Study of fibronectin expression in tumor-cells by dot-blot and in situ hybridization - quantitative-evaluation by image-analysis. Cell Biol. Int. Rep. 14, 701-715. Morrin, M. and Ward, O.P. (1989) Studies on interaction of Carbopol-934 with the hyphae of Rhizopus arrhizus. Mycol. Res. 92, 265-272. Morris, R. (1990) Image processing on the Macintosh software - review. Computer 23, 103-107. van Noorden, C.J.F. (1990) In situ measurements of enzyme reactions. Microsc. Anal. 7, 13-17. Obert, M., Pfeifer, P. and Sernetz, M. (1990) Microbial growth patterns described by fractal geometry. J. Bacteriol. 172, 1180-1185. Ochi, K. (1986) Occurrence of the stringent response in Streptornyces species and its significance for the initiation of morphological and physiological differentiation. J. Gen. Microbiol. 132, 2621-2631. Omasa, K., Aiga, I., Larcher, W., Onoe, M. and Shimazaki, K.I. (1987) Image-analysis of chlorophyll fluorescence transients for diagnosing the photosynthetic system of attached leaves. Plant Physiol. 84, 748-752. Packer, H.L. and Lilly, M.D. (UCL i) (1991) Improvements in, or relating to, biomass measurement. 21.03.91 U.K. Patent Appl. No. 9106007.9. Packer, H.L. and Thomas, C.R. (1990) Morphological measurements on filamentous microorganisms by fully-automatic image-analysis. Biotechnol. Bioeng. 35, 870-871. Paradiso, A.M., Machen, T.E. and Tsien, R.Y. (1987) Digital image-processing of intracellular pH in gastric oxyntic and chief cells. Nature 325, 447-450. Parry, W.L. and Hemstreet, G.P. (1988) Cancer-detection by quantitative fluorescence image-analysis. J. Urol. 139, 270-274. Pover, P. (1990) Introductory image analysis for microscopists. Image Enhance. Anal. Oct., p. 7-9. Preston, K. (1988) The need for standards in image-processing. Nature 333, 611-612. Pullman, J. and Bur, M. (1989) Immunoperoxidase stain quantitation - assaying reproducibility using computer-aided image-analysis. Lab. Invest. 60, A74. Reichl, U., Yang, H., Gilles, E.D. and Wolf, H. (1990a) An improved method for measuring the interseptal spacing in hyphae of Streptomyces tendae by fluorescence microscopy coupled with image-processing. FEMS Microbiol. Lett. 67, 207-210. Reichl, U., Buschulte, T.K. and Gilles, E.D. (1990b) Study of the early growth and branching of Streptornyces tendae by means of an image-processing system. J. Microsc. 158, 55-62. Saltzman, W.M., Langer, R. and Pasternak, S.H. (1987) Quantitative image-analysis for developing microstructural descriptions of heterogeneous materials. Chem. Eng. Sci. 42, 1989-2004. Sanchez, L., Regh, M., Biesterfeld, S., Chatelain, R. and Bocking, A. (1990) Performance of a TV image-analysis system as a microdensitometer. Analyt. Quant. Cytol. Histol. 12, 279-284. Sasov, A.V. (1989) An integrated PC-based image-analysis system for microtomography and quantitative-analysis of inner micro-object structure. J. Microsc. 156, 91-103.

18 Sato, K. and Shibuya, T. (1990) Quantitative image-analysis of fluorescent biogenic-amines in tissue. Eur. J. Pharmacol. 183, 421-422. Schleicher, A. and Zilles, K. (1990) A quantitative approach to cytoarchitectonics: analysis of structural inhomogeneities in nervous tissue using an image analyzer. J. Microsc. 157, 367-381. Shakespeare, A.P. and Verran, J. (1988) The use of automated image-analysis for rapid measurement of the in vitro attachment of Candida albicans to transparent acrylic. Lett. Appl. Microbiol. 6, 79-83. Simmons, J.A. and Richards, S.R. (1988) The volumes of rat and human lung cells as measured by an image analyser. Clin. Phys. Physiol. Meas. 9, 363-369. Singh, A., Yu, F.P. and McFeters, G.A. (1990) Rapid detection of chlorine-induced bacterial injury by the direct viable count method using image-analysis. Appl. Env. Microbiol. 56, 389-394. Sjollema, J., Busscher, H.J. and Weerkamp, A.H. (1989) Real-time enumeration of adhering microorganisms in a parallel plate flow cell using automated image-analysis. J. Microbiol. Methods 9, 73-78. Slocum, H.K., Malmberg, M., Greco, W.R., Parsons, J.C. and Rustum, Y.M. (1990) The determination of growth-rates of individual colonies in agarose using high-resolution automated image-analysis. Cytometry 11,793-804. Smith, M.A.L., Spomer, L.A., Mcclelland, M.T. and Meyer, M.J. (1989) Non-invasive image-analysis evaluation of growth during plant micropropagation. Plant Cell. Tiss. Org. Cult. 19, 91-102. Sulston, J., Mallett, F., Durbin, R. and Horsnell, T. (1989) Image-analysis of restriction enzyme fingerprint autoradiograms. Comput. Appl. Biosci. 5, 101-106. Tanke, H.J. (1989) Does light microscopy have a future'.) J. Microsc. 155, 405-418. Tsuprun, V.L., Orlova, E.V. and Mesyanzhinova, I.V. (1989) Structure of the ATP-synthase studied by electron-microscopy and image-processing. FEBS Lett. 244, 279-282. Twidwell, E.K., Patterson, J.A., Cherney, J.H., Bracket, C.E. and Johnson, K.D. (1989) Plant-tissue degradation measurement using image-analysis. Agronomy J. 81,837-840. Vannier, M.W., Biondetti, P.R., Butterfield, R.E. Jordan, D.M., Murphy, W.A. and Rickman, D.L. (1987) Multispectral magnetic-resonance image-analysis. CRC Crit. Rev. Biomed. Eng. 15, 117. Vecht-Lifshitz, S.E., Magdassi, S. and Braun, S. (1989) Effects of surface active agents on pellet formation in submerged fermentations of Streptomyces tendae. J. Disp. Sci. Technol. 10, 265-275. Vecht-Lifshitz, S.E., Magdassi, S. and Braun, S. (1990) Pellet formation and cellular aggregation in Streptomyces tendae. Biotechnol. Bioeng. 35, 890-896. Vivier, H., Portala, J.F. and Ports, M.N. (1989) Study of microporous membrane-structure by imageanalysis. J. Membr. Sci. 46, 81-91. Vonck, J. and Vanbruggen, E.R.J. (1990) Electron-microscopy and image-analysis of alcohol oxidase from yeasts. Ultramicrosc. 31,485. Wettlaufer, S.H. and Weinstein, L.H. (1988) Quantitation of polyamines using thin-layer chromatography and image-analysis. J. Chromatogr. 441,361-366. Wiebe, M.G., Robson, G.D. and Trinci, A.P.J. (1990) Edifenphos (Hinosan) reduces hyphal extension, hyphal growth unit length and phosphatidylcholine content of Fusarium grainearum A3/5, but has no effect on specific growth rate. J. Gen. Microbiol. 136, 979-984. Wied, G.L., Dytch, H.E., Bartels, P.H., Bibbo, M. and Dytch, H.E. (1989) Image-analysis in quantitative cyto-pathology and histopathology. Human Pathol. 20, 549-571. Yu, X. and Egelman, E.H. (1990) Image-analysis reveals that Escherichia-coli reca protein consists of 2 domains. Biophys. J. 57, 555-566. Zygourakis, K. and Glass, M.W. (1988) Macropore size analysis using digital image-processing and a stereological model. Chem. Eng. Comm. 70, 39-55. Zyrd, J.P., Bodenmann, J. and Burki, M. (1987) Plant-cell protoplasts - population-pattern analysis using image-processing - preliminary-results. Experientia 43, 665.