The measurement of body fat distribution using somatotype photographs and computer-assisted imaging techniques.

A computer-assisted image analysis technique has been devised which allows the measurement of body fat distribution from archived somatotype photographs. This report describes the technique, which enables useful measurements to be extracted from somatotype photographs which were taken over 50 years ago. This paper describes the process of digitization of photographs, the measurement of body diameters using computer-assisted edge detection, the quantitative analysis of measurement error in this procedure, and validation techniques employed.

An augmented computer model of motor unit reorganization in neurogenic diseases of skeletal muscle.

A computer model of denervation and complete reinnervation in skeletal muscle was originally developed for the purpose of furthering an understanding of the underlying mechanisms of motor unit reorganization in neurogenic diseases. We now describe its successor, a computer model for investigating different rates of denervation and reinnervation, as well as incomplete reinnervation. The new model introduces the concept of permanent denervation and features enhanced interactive control over the distribution of motor unit centers and additional measures of dispersion and co-dispersion of muscle fibers. The use of this model for investigating pathophysiologically significant issues in denervating diseases is illustrated with five different sets of parameters. These simulate some of the processes that may be operational in chronic spinal muscular atrophy, amyotrophic lateral sclerosis, and progressive postpolio muscular dystrophy. The enhanced model will allow in-depth analysis of the influence of hypothesized pathophysiological processes on clinical, electrophysiological and pathological outcomes in human disease.

Quantitative morphometric study of muscle in inclusion body myositis.

Clinical and electromyographic findings do not clearly distinguish inclusion body myositis (IBM) from chronic polymyositis (PM). The rimmed vacuoles and filamentous nuclear and cytoplasmic inclusions that characterize IBM are often sparse and may be overlooked; conversely, these features may occasionally be seen in other diseases. Preliminary studies suggested that muscle fiber hypertrophy occurred more frequently in IBM than in PM. To investigate whether fiber hypertrophy can be used to improve the ability to separate IBM from PM, we report a morphometric analysis of 28 IBM cases, 22 PM and 22 dermatomyositis (DM) cases. The analysis, using a computer automated system, included proportion of hypertrophied fibers and also fiber type proportions, average fiber diameter, proportion of atrophic and angulated fibers, and the co-dispersion index (CDI). The proportion of hypertrophied fibers was greater in IBM than the other two conditions (IBM (mean +/- SEM) 31.0 +/- 4.7% and 12.2 +/- 2.4% for type 1 and type 2 fibers, respectively, compared to 9.8 +/- 3.0% and 3.3 +/- 1.7% in PM, and 7.7 +/- 2.7% and 3.9 +/- 1.9% in DM). These differences were statistically significant (P < 0.05) in both sexes for type 1 fibers and in women for type 2 fibers. Also, the average fiber size and hypertrophy factors for type 1 and type 2 fibers were increased in IBM compared to PM and DM. This study confirms that the presence of muscle fiber hypertrophy in biopsies from IBM patients may help differentiate them from other clinically similar inflammatory myopathies.

Contextual analysis and intermediate cell markers enhance high-resolution cell image analysis for automated cervical smear diagnosis.

Until now, efforts to automate cervical smear diagnosis have focused on analyzing features of individual cells. In a complex specimen such as that obtained from a cervical scrape, diagnostically significant cells may not be adequately represented or may elude detection by the automated technology. An approach is needed that extracts additional quantitative information from cervical smears beyond what the cell-by-cell approach can provide. A new methodology, contextual analysis, was developed to extract global quantitative information about cells, cell clusters, and background debris. This pilot study was designed to compare the efficacy of contextual analysis with high-resolution, single cell analysis and the analysis of intermediate cell markers. Thirty-four samples prepared as monolayers and stained with the Feulgen-Thionin/Congo Red stain were measured. Contextual analysis alone was able to classify 91% of the smears correctly; single cell analysis classified 94% of the cells correctly; and the intermediate cell analysis correctly identified the smear diagnosis for 84% of the cells. When all three analysis methods were combined into a simple smear level classifier, the overall smear classification accuracy was improved over those obtained using the three methodologies alone.

Monte Carlo based dosimetry and treatment planning for neutron capture therapy of brain tumors.

Monte Carlo based dosimetry and computer-aided treatment planning for neutron capture therapy have been developed to provide the necessary link between physical dosimetric measurements performed on the MITR-II epithermal-neutron beams and the need of the radiation oncologist to synthesize large amounts of dosimetric data into a clinically meaningful treatment plan for each individual patient. Monte Carlo simulation has been employed to characterize the spatial dose distributions within a skull/brain model irradiated by an epithermal-neutron beam designed for neutron capture therapy applications. The geometry and elemental composition employed for the mathematical skull/brain model and the neutron and photon fluence-to-dose conversion formalism are presented. A treatment planning program, NCTPLAN, developed specifically for neutron capture therapy, is described. Examples are presented illustrating both one and two-dimensional dose distributions obtainable within the brain with an experimental epithermal-neutron beam, together with beam quality and treatment plan efficacy criteria which have been formulated for neutron capture therapy. The incorporation of three-dimensional computed tomographic image data into the treatment planning procedure is illustrated. The experimental epithermal-neutron beam has a maximum usable circular diameter of 20 cm, and with 30 ppm of B-10 in tumor and 3 ppm of B-10 in blood, it produces (with RBE weighting) a beam-axis advantage depth of 7.4 cm, a beam-axis advantage ratio of 1.83, a global advantage ratio of 1.70, and an advantage depth RBE-dose rate to tumor of 20.6 RBE-cGy/min (cJ/kg-min). These characteristics make this beam well suited for clinical applications, enabling an RBE-dose of 2,000 RBE-cGy/min (cJ/kg-min) to be delivered to tumor at brain midline in six fractions with a treatment time of approximately 16 minutes per fraction. With parallel-opposed lateral irradiation, the planar advantage depth contour for this beam (with the B-10 distribution defined above) encompasses nearly the whole brain. Experimental calibration techniques for the conversion of normalized to absolute treatment plans are described.

A computer model of denervation-reinnervation in skeletal muscle.

A computer model of the process of denervation and complete reinnervation of skeletal muscle has been developed for the purpose of exploring underlying mechanisms and for use in the development of new clinical and research tools for evaluating neuromuscular disease. Progressive motor neuron death and reinnervation in this model reproduces the fiber-type grouping, increased fiber density, and minimal increase of motor unit size seen in human chronic denervating diseases. Studies using the model suggest that (1) preferential involvement of motor units of one type could account for the abnormal fiber-type proportions observed in some diseases, (2) reinnervation by axons innervating adjacent fibers is compatible with single fiber multielectrode study results in that it does not produce a large increase in motor unit area, and (3) such reinnervation is sufficient to account for the increases in motor unit density that have been observed. The model has also been used in the development and testing of the Codispersion Index, a measure of the codistribution of two fiber types, which is useful in detecting fiber-type grouping.

Evaluation of contextual analysis for computer classification of cervical smears.

A procedure for automated analysis of cervical smears has been implemented in an image cytometry system. Smears are described exclusively in terms of global and contextual information extracted by pattern-recognition algorithms and represented by a vector of proportions of cellular object types. Linear discriminant functions, based on a Fisher criterion, are derived to classify smears with a cross-section of diagnoses into two broad categories, normal and abnormal. Results obtained from 83 smears indicate 78% correct classification. In contrast to most automated systems, good classification results were obtained in normal smears with benign changes caused by inflammation and with postmenopausal atrophia and in abnormals with mild dysplasia. These findings suggest that contextual analysis may be sensitive to subtle changes in cellular morphology and to progressive patterns of dysplasia. When used with standard isolated cell analysis, contextual analysis may provide additional complementary information for automated cervical prescreening.

Automated system for the analysis of muscle and nerve biopsies.

An automated system has been developed for morphometric analysis of muscle and nerve fibers as seen in routine biopsy tissue preparations. This system, which is suitable both for research and clinical applications, is now in routine use at New England Medical Center Hospital. It is designed around commonly available hardware and uses our own custom software. Special sample preparation is not required, and the system will process almost any biopsy that is free from obvious artifact and suitable for manual assessment. In a typical biopsy, the system provides nearly full automation for the edge (boundary) determination of over 90%of the fibers encountered. An efficient, interactive mode allows for rapid analysis of the remaining 10%. Once fiber boundaries are determined, the system provides fully automatic measurement of the usual parameters such as fiber diameter, as well as a number of parameters not routinely measured with existing manual techniques. Typical analysis times are 35 min for muscle and 2 h for nerve. Complete report generation is provided, with hardcopy available in the pathologist's office. This paper describes the operational aspects of the use of the automated analysis system in routine biopsy work.

Spectral bandwidth in automated leukocyte classification.

Investigators have repeatedly pointed out the importance of spectral information in the automated classification of white blood cells. In general, monochromatic images recorded through two or three color filters are used to extract this information. Although it has generally been thought that the use of narrow band filters provides "cleaner" color information than is obtainable through wide band filters, the choice has not been fully investigated and the question is far from being settled. The use of wide band filters has the clear practical advantage of increased light levels at the detector, resulting in higher signal-to-noise ratio with less demand on light source design. In order to investigate this issue, a series of 681 leukocytes of the most frequently occurring types were digitized by the use of both narrow (10 nm) and wide (90 nm) band filters. Parameters were extracted independently from both sets of images. These parameters were then used to develop a classifier for each set of images. The choice of features and classifier results indicate that there are no major performance differences between the two types of filters.

Detecting infection-related changes in peripheral blood smears with image analysis techniques.

High-resolution image analysis has the potential to flag subtle changes in white blood cell morphology that may indicate the presence of certain diseases. A study was made of the feasibility of identifying patients with hematologic bacterial infections (sepsis) using measurements on Wright-Giemsa-stained peripheral blood smears. Neutrophils and lymphocytes from a group of patients with sepsis and from a control group were digitized, and parameters quantifying geometry, color, texture and shape were extracted. While color parameters differed the most between the infected and control samples, substantial differences in geometric, texture and shape parameters also were observed. Analysis of the data showed that individual neutrophils and lymphocytes from patients with sepsis were distinguishable from those of the control group with better than 84% accuracy. When average parameters were calculated from all cells of one type for each specimen, 100% accurate classification was obtained. These studies demonstrate that the image-analysis techniques used are sensitive enough to detect disease-related changes in cell morphology that are generally too subtle for reliable detection by the human eye. Future experiments will determine the specificity of this test for bacterial infections and will explore the possibility of using image analysis techniques on peripheral blood to detect and monitor a wide variety of diseases.

The co-dispersion index for the measurement of fiber type distribution patterns.

There is a need for a practical and statistically manageable quantitative index of fiber type co-dispersion in skeletal muscle to allow investigation of the distribution of motor units in normal muscle and their rearrangement in pathological states. A new measure for this purpose, the Co-Dispersion Index (CDI), is introduced. This measure is based on contingency table analysis of nearest-neighbor relationships between muscle fibers. The CDI has a continuous range of values from -1 to +1, in which larger negative values indicate a greater tendency toward regular intermixing of fibers of different types, and larger positive values a greater segregation of fiber types. CDI evaluation overcomes the limitations of previously published quantitative methods for co-dispersion measurement and is well correlated with subjective human estimates of fiber type grouping.

Computer image analysis of variance between human chromosome replication sequences and G-bands.

A computer image analysis system was applied to the quantitative study of chromosomal early- and late-replication patterns from the leukocytes of several normal human donors, and these patterns were compared with the chromosomal G-banding patterns. The first and last few hours of replication were discriminated by selective bromodeoxyuridine vs. thymidine incorporation in DNA and a Hoechst-blacklight-Giemsa stain technique. Image analysis with Tufts Piquant system involved automatic determination of chromosome boundaries, centromeres and telomeres, linear chromatid axes, chromatid density measurements along each axis, and comparative length normalized density profiles for each chromatid and the chromosome. Consistent complementary early- and late-replication patterns were determined for autosomes 1-6 and the X chromosomes. Limited intracellular or interindividual variability occurred in the intensity of a few active replication peaks but not in their location. However, there were very distinct regions of noncorrespondence between the late-replication patterns and the G-band patterns, in contrast with previous observations, although many similarities were also evident. These differences are interpreted with reference to a general model of replication sequence control of cell differentiation.

Discrimination and consistency of five myosin ATPase stains in human normal and Duchenne dystrophic muscle.

Limitations in the ability of the human visual system to assess accurately the relative staining densities of individual fibers in muscle tissue stained for myosin. ATPase can complicate the objective evaluation of fiber type populations. In this study a novel approach is employed which utilizes human visual capabilities to provide accurate fiber classification. Using this approach, the ability of five ATPase staining techniques to discriminate fiber type categories in single samples of human normal and Duchenne dystrophic skeletal muscle is evaluated, as is the consistency of the fiber type classifications between stains. While no major discrepancies in fiber typing were observed in the sample of normal muscle, significant differences in classification, along with a decrease in the ability to discriminate fiber types were noted in the sample of Duchenne muscular dystrophy. For the most part, these discrepancies were resolved by a re-interpretation of the staining characteristics of fibers in one stain.

Automated morphometric analysis of corneal endothelial cells.

This paper describes a feasibility study for computer image analysis of corneal endothelial specular micrographs. The analysis involved image enhancement, boundary detection of individual cells, and subsequent calculation of cell area, cell area distribution, cell perimeter, cell orientation, and cell diameter. Cell areas produced by automated analysis agreed well with manually measured cell areas from the same photomicrographs. The computer methods required little human guidance or correction, produced accurate results, and could be made to run fairly quickly with modest computing resources.

Scene segmentation techniques for the analysis of routine bone marrow smears from acute lymphoblastic leukemia patients.

Automated analysis of lymphoblast cell morphology is being evaluated as a basis for predicting the response to therapy of patients with acute lymphoblastic leukemia. A new technique of scene segmentation particularly applicable to the "cluttered" images of cells in routine bone marrow smears is described. Morphologic characteristics of lymphoblasts found in bone marrow smears made at time of diagnosis were measured by an automated, interactive image-processing system using the new scene segmentation technique. These characteristics, on a patient by patient basis, are being compared to remission length and survival data to develop and test new prognostic methods.