H. M. Aus

Computer analysis of chromatin arrangement and nuclear texture in follicular thyroid tumours

SummaryThe differentiation of the thyroid glands follicular neoplasias into adenomas and carcinomas is currently done using the histological criteria recommended by WHO. This pilot study of 10 human follicular carcinomas and 10 folliculars adenomas demonstrates the possibility of a cytological classification using digital picture processing of high resolution cell images. Giemsa stained paraplast sections were scanned with a Colour-TV-camera, different channels were used with respect to staining and analyzing methods and computed with an image processing system. The computer aided cytophotometric methods detected significant differences in the chromatin arrangement and structure.

Automated quantitative analysis of single and double label autoradiographs.

A method for the analysis of silver grain content in both single and double label autoradiographs is presented. The total grain area is calculated by counting the number of pixels at which the recorded light intensity in transmission dark field illumination exceeds a selected threshold. The calibration tests included autoradiographs with low (3H-thymidin) and high (3H-desoxyuridin) silver grain density. The results are proportional to the customary visual grain count. For the range of visibly countable grain densities in single labeled specimens, the correlation coefficient between the computed values and the visual grain counts is better than 0.96. In the first emulsion of the two emulsion layer autoradiographs of double labeled specimens (3H-14C-thymidin) the correlation coefficient is 0.919 and 0.906. The method provides a statistical correction for the background grains not due to the isotope. The possibility to record 14C tracks by shifting the focus through the second emulsion of the double labeled specimens is also demonstrated. The reported technique is essentially independent of size, shape and density of the grains.

Computer aided analysis of chromatin network and basophil color for differentiation of mononuclear peripheral blood cells.

Computer aided differentiation of plasmoblasts, Pfeiffer cells, immunoblasts, lymphocytes and centrocytes is achieved with the parameters of chromatin network arrangement and structure, and multispectral cytoplasm color. The digital methods involve: (a) segmenting the nuclear image into topographic sections and analyzing the optical density distribution from the chromatin in these sections; (b) determining the nuclear structure with a 7 x 7 median filter, gradient filter and contour following algorithms; and (c) clustering two-dimensional chromatic data from panoptically stained cellular components. The parameters reported here are a subset of those needed for the automated diagnosis of many hematologic diseases especially the leukemias.

Statistical evaluation of computer extracted blood cell features for screening populations to detect leukemias

This paper outlines image segmentation, feature extraction and classification methods which screen and diagnose blood malignancies. New algorithms had to be developed because analyzing blood malignancies requires higher scanning densities and higher optical magnification than used in screening normal blood cells. The cell image segmentation method combines color differences, equidistance isograms, geometric operations, and a cell model. The algorithm always starts with the largest color differences and successively detects less certain areas. This eliminates the need for contour following algorithms. The feature extraction combines geometric parameters with texture and color. “Classification And Regression Tree” statistical software test the classification power of the cell markers extracted by the image processing. The feature distributions from the tested blood cell population correlate directly to the specific blood malignancies.

Lysosomal Enzyme Activity in Poliovirus-infected HeLa Cells and Vesicular Stomatitis Virus-infected L Cells: biochemical and Histochemical Comparative Analysis with Computer-aided Techniques

Summary The behaviour of lysosomal enzymes in poliovirus-infected HeLa S3 cells and vesicular stomatitis virus (VSV)-infected L cells was investigated both biochemically using enzyme assays, and histochemically using acid phosphatase dependent staining. The presence of the enzyme was shown histochemically under the light microscope by its reaction with naphthole-AS-BI-phosphate and a coupling reaction with diazotized pararosaniline. The light absorption of this stain in infected and uninfected cells was measured on a Universal Micro Spectrophotometer (UMSP-1) and recorded on-line as gray value cell images in a PDP-12 computer. These scanned images were analyzed by FORTRAN programs on a UNIVAC1108. The histochemically obtained distributions of the lysosomal enzyme are comparable to the results of the biochemical analysis. Lysosomes of poliovirus-infected cells displayed a release of lysosomal enzymes into the cytoplasm starting at 3 h after infection; VSV infection did not produce this type of effect. This investigation shows that it is possible to extract and demonstrate specific virus dependent changes using computer-aided cytophotometric techniques.

TECHNIQUES APPLICABLE TO COMPUTER-AIDED CYTOPHOTOMETRY IN VIROLOGY

Computer-aided cytophotometry is a new tool in the analysis of virus-infected cells. Structural changes of the cell morphology can easily be recognized and interpreted by the analytic systems available. More specific biochemical events that occur during virus replication can in certain instances be recorded and recognized if well defined trace markers are applied. This is demonstrated on poliomyelitis virus-infected HeLa cells.

Cytophotometric analysis of lytically and persistently infected tissue culture cells with measles virus

Abstract Lytically and persistently infected VERO and LUB cells were cytophotometrically characterized with the aid of computer analysis. Images scanned at 260 and 280 nm were processed by computer algorithms. With this approach, infected cells could be segmented and differentiated from uninfected cells. Lytically and persistently infected cells could be distinguished by distinct differences in nuclear and cytoplasmic optical densities. These findings are supported by biologic data based on the analysis of virus‐specific proteins and nucleic acids. The applied computer aided cytophotometry provides a new approach in the study of virus‐cell interaction.

High resolution image analysis.

The recent review article by Dr. Preston on “High Resolution Image Analysis” (8) confronts the reader with the dismal and old state of imaging cytometry. Although the author summarizes the past correctly, his analysis of the present situation is inaccurate. Contrary to the title of the paper and the subsection”High-Resolution Microscopy in Hematology,” none of the systems he describes can be categorized as “high spatial resolution.” All the microscope image acquisition systems reviewed in his paper scan with a scanning density considerably lower than 10 pixels per sam. This choice of lowresolution scanning is based, among other things, on reducing the amount of data to be processed; but this often contradicts the requirements posed by the clinical applications. Earlier work by Rzeszotarski (9) showed that the scanning density used in imaging cytometry is too low for detecting all the visible information in the light microscope. As documented in our publications (2, 6), high-resolution scanning and image acquisition starts above 10-12 pixels per xm. Below this limit, a substantial amount of indispensable information content in a microscopic image is lost. Anything less is simply not high resolution. In hematology, automation efforts have all concentrated on saving manpower and on screening normal cells faster than can be done by the best trained laboratory personnel. But the more important problem is to provide reliable information about blood cells which indicates specific disorders. Commercial systems have not taken this approach. Table 1 in Dr. Preston’s review article demonstrates the problem using early results from Geometric Data’s Hematnak white blood cell counting (WBC) system. The normal white blood cells can typically be classified with a 3% false-alarm rate. The rare cell types (in ‘Ihble 1: myelocytes, promyelocytes, nucleated red cells, blasts, and plasmacytes) are, however, seldom classified correctly and are mostly collected in the catch-all cell-type “OTHER.” The false identification of a blast cell as either an cosinophil or basophil is particularly distressing to the hematologist. Both of the latter cell types have distinct cytoplasm granulation and color which set them clearly apart from the blasts. Moreover, the monocyte/neutrophil confusion rate (3%) is also unacceptable in clinical routine. These cell types are morphologically so distinct that no laboratory technologist would confuse them. Underscanning is one rcason for such mistakes (5). Commercial differential WBC systems have been designed to screen for normal cells. But blood samples can also contain other types of cells. Because these cells occur so infrequently in a routinely screened populaiion, the WBC system designers have conveniently defined them as “rare cells” and have ignored their clinical importance; precisely these “rare cells” are, however, key indicators of a hematological disorder. The rare cells are usually not so easily identified, and typically appear only in conjunction with some disorder; moreover, the sizes of the discriminating features are often not much greaten than the maximum resolution limits of the microscope optics (6, 7). Aside from the financial aspects, the poor clinical acceptance of the white blood cell machines can be attributed largely to the lack of solid diagnostic data. Nonetheless, health care delivery requires a system that not only screens normal cells but also reliably diagnoses and correctly classifies rare cells from blood disorders in the same blood smears routinely used to screen all patients. But image processing and “true” high-resolution microscopy can do more than Dr. Preston asserts in his review article! One goal of our project over the last years has been to investigate and develop new methods for extracting clinically relevant diagnostic information about leukemias from Pappenheim-stained blood smears. Fur years the hematologists have contended that there exist small differences in the fine structured texture ofso-called blast cells, according to each leukemia. To the best of our knowledge, we have been one of the first groups to reliably measure and directly analyze the blasts (1, 3, 4). The occurrence of a single blast cell in the patient’s peripheral blood is a first indicator that a leukemia is present. In health-care delivery, it is mandatory to detect this blast cell as early as possible. With respect to leukemia, the diagnosis is strongly related to exactly identifying each white blood cell occurring in the peripheral blood. The articles listed below should, we hope, provide the interested reader with a more encouraging outlook for “real” highresolution microscopy than Dr. Preston does.

Automatisierte Chromosomenanalyse Mittels eines Hochauflösenden TV-Mikroskops

Seit Jahren bemuhen sich verschiedene Arbeitsgruppen, die Arbeit im genetischen Labor durch eine zytophotometrische Erfassung der Metaphasen und ihre anschliesende Auswertung durch den Rechner zu unterstutzen. In dieser Arbeit wird nun ein Bildverarbeitungssystem beschrieben, das mit Hilfe der beiden Parameter “Chromosomengrose” und “ Zentromer index” ungebanderte Chromosomen einer Metaphase automatisch zu einem Karyotyp anordnet. Es dient im Moment als Basis fur ein Programmsystem, das die Sequenz der Zentromerteilung bei den Chromosomen von Muntjak Zellen bestimmt /3/.

Computeranalyse von Elektronenmikroskopischen Bildern

Bislang war die Bildqualitat bei der TV-Ubertragung vom Transmissionselektronenmikroskop (TEM) sehr mangelhaft. Niedrige Lichtenergie aus dem TEM, fehlende Empfindlichkeit und Rauschen im Bildempfanger waren die Hauptschwierigkeiten. Die Bearbeitung der elektronenmikroskopischen Bilder erfolgte deshalb vorwiegend auf photometrischem Wege an Mikrografien oder uber Eingaben auf ein Digitalisiertablett interaktiver Bildanalysesysteme an aus dem TEM herausgespiegelten Bildern. Erst in jungster Zeit ist mit der Entwicklung hochempfindlicher Kameras, schneller Analog-Digital Wandler und groserer Laborrechnerkapazitat eine direkte Bilderfassung uberhaupt moglich geworden.