Automatic image segmentation of nuclear stained breast tissue sections using color active contour model and an improved watershed method

Automatic image segmentation of immunohistologically stained breast tissue sections helps pathologists to discover the cancer disease earlier. The detection of the real number of cancer nuclei in the image is a very tedious and time consuming task. Segmentation of cancer nuclei, especially touching nuclei, presents many difficulties to separate them by traditional segmentation algorithms. This paper presents a new automatic scheme to perform both classification of breast stained nuclei and segmentation of touching nuclei in order to get the total number of cancer nuclei in each class. Firstly, a modified geometric active contour model is used for multiple contour detection of positive and negative nuclear staining in the microscopic image. Secondly, a touching nuclei method based on watershed algorithm and concave vertex graph is proposed to perform accurate quantification of the different stains. Finally, benign nuclei are identified by their morphological features and they are removed automatically from the segmented image for positive cancer nuclei assessment. The proposed classification and segmentation schemes are tested on two datasets of breast cancer cell images containing different level of malignancy. The experimental results show the superiority of the proposed methods when compared with other existing classification and segmentation methods. On the complete image database, the segmentation accuracy in term of cancer nuclei number is over than 97%, reaching an improvement of 3–4% over earlier methods.     

Studies on ribonucleic acid and homopolyribonucleotide formation in neuronal, glial and liver nuclei

1. Various types of nuclear preparations, with different ratios of neuronal to glial nuclei, were isolated from guinea-pig cerebral grey matter and ox cerebral grey matter and white matter. Conditions appropriate for the separate assay of RNA and poly A formation were described. Comparative rates of RNA and poly A formation were studied in cerebral and liver nuclei. 2. RNA polymerase activity per nucleus is higher in neuronal nuclei than in glial nuclei. In liver nuclei, the activity is much lower than in cerebral nuclei. The physical relationship between RNA polymerase and deoxyribonucleoprotein seems to differ in neuronal, glial and liver nuclei. 3. Poly A polymerase activity in liver nuclei is selectively activated by Mn(2+) and inhibited by GTP, CTP and UTP. On a DNA basis, the activity in an aggregate enzyme is the same as in intact nuclei. Poly A polymerase activity per nucleus is much higher in liver nuclei than in neuronal nuclei. Glial nuclei show an intermediate activity. 4. It is suggested that, in neuronal nuclei, the synthesis of RNA is more prominent than that of poly A under conditions where both polymers are formed simultaneously. This contrasts with liver nuclei, where more poly A is made than RNA. 5. In neuronal nuclei, the rate of CTP incorporation is much higher than in glial and liver nuclei. This incorporation is most probably due to poly C synthesis.     

On the association of centrioles with the interphase nucleus.

Preparations of nuclei from rat liver and bovine spleen purified by centrifugation through dense sucrose solutions are shown to contain centrioles. These centrioles retain their in situ ultrastructure and are surrounded by a network of filaments adjacent to the nucleus and probably attached to it. The number of centrioles in isolated nuclei depends on the conditions of cell homogenization. Under certain conditions of homogenization, the fraction of purified nuclei contains almost all centrioles of the original tissue. The number of centrioles in isolated nuclei sharply decreases if the nuclei are rehomogenized under conditions that do not cause damage to nuclei. The number of nucleus-associated centrioles does not decrease after solubilization of nuclear membranes by Triton X-100. Nuclei retain the associated centrioles after treatmentwith RNase-free DNase I. It is concluded that in interphase the centrioles are associated with the nucleus and that this association which is probably mediated by filaments involves nuclear structures other than nuclear membranes or whole chromatin.     

Physical characteristics of mouse sperm nuclei.

The nuclei of epididymal sperm, isolated from C57BL/6J and CBA/J inbred mice by their resistance to trypsin digestion, retain the shape differences of the intact sperm head. Various physical characteristics of these nuclei were measured and compared. The measurement of the projected dimensions of nuclei showed that the CBA nuclei are 13.5% longer than C57BL/6 nuclei (8.64 +/- 0.02 mum compared with 7.61 +/- 0.02 mum), 0.8% narrower (3.51 +/- 0.01 vs. 3.54 +/-0.01 mum) with 6.8% more area (22.34 +/- 0.10 vs. 20.91 +/- 0.09 mum2). However, the volumes of the nuclei as based on reconstructing calibrated electronmicrographs of serial sections of the nuclei indicated that CBA are about 7% smaller than C57BL/6 nuclei (3.72 +/- 0.08 vs. 4.01 +/- 0.03 mum3). The buoyant density of the CBA nuclei is 1.435 +/- 0.002 g/cm3 compared with 1.433 +/- 0.002 g/cm3 for the C57BL/6 nuclei as determined on linear CsCl and Renografin-76 density gradients and confirmed by a technique utilizing physiological tonicities. Therefore, the average mass of the CBA nuclei is less than that of the C57BL/6 nuclei (5.34 +/- 0.12 vs. 5.75 +/- 0.05 pg). The sedimentation velocities at unit gravity of nuclei from 11 inbred strains differ over a range of more than 6% with CBA nuclei sedimenting about 2.0% more slowly than C57BL/6 nuclei. We show that for these nuclei the sedimentation velocity can be related to their buoyant density, volume and a sedimentation shape factor. Within the errors of our measurements of these various characteristics, it was found that C57BL/6 and CBA nuclei have similar sedimentation shape factors. Therefore, the difference in sedimentation velocity between these nuclei appears to be primarily a result of differences in volume. The possible applications of these techniques to the physical separation of sperm are evaluated in the discussion.     

In Vitro RNA Synthesis by Rat Liver and I Leukemic Nuclei. Isolation of Undegraded RNA

Incubation of nuclei from rat liver or human leukemic cells in the presence of ³H-UTP² and other factors results in th incorporation of label into a material precipitable by acid, alcohol or ether. This materials is isolated by phenolsds extraction, is sensititve to ribonuclease digestion and presumed to be RNA.

The addition of Cu++ to the incubation system is necessary to inhibit RNA breakdown and allows the isolation of undegraded RNA without interefering with th incorporation of radiosactivity. The time patterns of labl incorporation by the two nuclei preparations are different. Whereas label incorporation by th two nuclei preparations are different. Whereas labelincorporation by liver nuclei continues to increase up to 60 minutes, incorporation by th leukemic nuclei is high during the first 10 minutes and continues at a slower rate up to 45 minutes of incubation. further, th two nuclei preparations also synthesize diferent RNA species. While liver nuclei synthesize RNA sedimenting at 4.5S and 7S to 13S, leukemic nuclei synthesize a heterogeneous, polydisperse type of RNA.     

核微管组织中心的研究

本文利用间接免疫荧光手段对间期核微管组织中心的性质进行了初步探讨。当分离核与6S微管蛋白保温后,用抗管蛋白抗体进行间接免疫荧光染色,在群体中只有极少数核上有微管生长。Triton X-100或溶血卵磷脂去膜核的实验证实能长微管的核是核膜缺失或破损核。用秋水仙素解聚核上长出的微管,可见核内荧光亮点数多于染色体数,说明核内微管组织中心数目多于着丝粒数,很可能包括着丝粒和一部分染色质颗粒。将秋水仙素和6S微管蛋白混合后与去膜核保温,结果6S微管蛋白与核内微管组织中心有亲和力。本文就这些结果进行了讨论。     

Dynamic changes in microtubule configuration correlate with nuclear migration in the preblastoderm Drosophila embryo

Drosophila embryogenesis is initiated by a series of syncytial mitotic divisions. The first nine of these divisions are internal, and are accompanied by two temporally distinct nuclear movements that lead to the formation of a syncytial blastoderm with a uniform monolayer of cortical nuclei. The first of these movements, which we term axial expansion, occurs during division cycles 4-6 and distributes nuclei in a hollow ellipsoid underlying the cortex. This is followed by cortical migration, during cycles 7-10, which places the nuclei in a uniform monolayer at the cortex. Here we report that these two movements differ in their geometry, velocity, cell-cycle dependence, and protein synthesis requirement. We therefore conclude that axial expansion and cortical migration are mechanistically distinct, amplifying a similar conclusion based on pharmacological data (Zalokar and Erk, 1976). We have examined microtubule organization during cortical migration and find that a network of interdigitating microtubules connects the migrating nuclei. These anti-parallel microtubule arrays are observed between migrating nuclei and yolk nuclei located deeper in the embryo. These arrays are present during nuclear movement but break down when the nuclei are not moving. We propose that cortical migration is driven by microtubule-dependent forces that repel adjacent nuclei, leading to an expansion of the nuclear ellipsoid established by axial expansion.     

Suppressive effect of protease inhibitors on heterokaryons containing chick erythrocyte nuclei

Nuclei of active cells (HeLa, mouse fibroblasts) partnered with chick erythrocyte nuclei in heterokaryons are suppressed, as judged by a decreased rate of ³H-uridine incorporation and diminished nuclear binding of ³H-actinomycin D. The extent to which active partner nuclei are suppressed, the extent to which erythrocyte nuclei are reactivated, and the degree of sensitivity of heterokaryons towards certain inhibitors of proteolytic enzymes, all correlate strongly with the ratios of erythrocyte nuclei to active nuclei. Thus, reactivation of individual erythrocyte nuclei is reduced progressively and active nuclei are suppressed progressively as the ratio of erythrocyte nuclei per active nucleus in heterokaryons increases. This erythrocyte nuclear-dose dependent suppression is markedly amplified when heterokaryons are grown in the presence of protease inhibitors. The protease inhibitors found to affect heterokaryons are low molecular weight (<400) inhibitors of trypsin-like enzymes: -1-tosylamide-2-leucyl chloromethyl ketone (TLCK), N-α-tosyl- -arginine methyl ester (TAME) and N-benzoyl- -arginine amide (BAA). They affect heterokaryons at concentrations comparable to the minimal concentrations at which they inhibit trypsin. Nonfused HeLa cells, mouse fibroblasts, or their homokaryons are refractory to protease inhibitors at these concentrations.Reactivation of chick erythrocyte nuclei in a heterokaryon may involve release of suppressors ordinarily confined to the erythrocyte nucleus, with subsequent redistribution of suppressor among all the nuclei of the heterokaryon. Under these circumstances the state of nuclear activity will depend on the quantity of suppressor per individual nucleus; within the erythrocyte nucleus the suppressors will decrease its rate of reactivation, when they migrate into an active nucleus they will suppress it. These suppressors, either in transit between the nuclei, or within the nuclei, may be hydrolysed by intracellular proteases.     

Effect of chromatin decondensation on the intranuclear matrix

We have studied the effect of chromatin condensation on the morphology of the residual structures isolated from rat liver nuclei. DNAse I digestion followed by high salt extraction of nuclei in the presence of Mg++ yields residual structures consisting of a dense peripheral layer surrounding an internal network, similar to those described by Berezney and Coffey [6]. These structures are stable at low ionic strength in the presence of EDTA. When nuclei swollen in EDTA are digested with DNAse II in the presence of EDTA, structures devoid of internal network are obtained even without subsequent treatment with high salt. When swollen nuclei are exposed to Mg++ a specific recondensation of chromatin takes place. The residual structures from recondensed nuclei are similar to those isolated from control nuclei in the presence of Mg++. The results suggest that the integrity and stability of the intranuclear matrix are acquired in the course of the isolation procedure and this is favoured by chromatin condensation.     

ISOLATION AND FRACTIONATION OF RAT BRAIN NUCLEI

A method for isolating pure and unaltered nuclei from rat brain by means of differential centrifugation is described. The isolated nuclei are further separated into discrete fractions of neuronal, astrocytic, and glial nuclei, with a yield amounting to 20 to 25% of the DNA of the original homogenate. Both the morphology and size of the nuclei remained unchanged. Problems concerning the composition of the isolation media, the use of detergents, as well as those raised by density gradient centrifugation in sucrose, Ficoll, and Dextran are discussed. Some values for the density of each type of brain nuclei are suggested.     

The double nucleus of the sertoli cell in the lizard Anolis carolinensis

Over 90% of the Sertoli cells in the testes of adult lizards (Anolis carolinensis) are binucleate. The nuclei occur in closely associated pairs in the basal cytoplasm of the Sertoli cells that line the testis tubules. The two nuclei of a pair are of similar volume, and each usually contains one conspicuous rounded nucleolus. The average volume of individual nuclei varies from 367.8 mum(3) in spermatogenically active testes in March to 172.5 mum(3) in September, when testes are regressed. The irregular shape of the Sertoli nuclei is particularly pronounced during testicular regression. Until initiation of spermatogenesis in hatchling lizards, Sertoli cells have a single nucleus containing patches of hetcrochromatin. With the appearance of prophase stages of primary spermatocytes, a few paired Sertoli nuclei can be found, and the nuclei increasingly exhibit the homogeneous euchromatic nucleoplasm of the adult. The average volume of individual nuclei in lizards under 4 months of age is less than a third the volume of Sertoli nuclei in reproductivcly active adults. The appearance of binucleate cells at this time documents a doubling of the amount of desoxynucleic acid in Sertoli cells preparatory to their growth and expanded functions during spermatogenesis.     

Nuclear death in living Tetrahymena: the case of the haploid nuclei

During sexual conjugation in Tetrahymena the micronucleus divides meiotically, producing four haploid nuclei. While one of these nuclei divides mitotically to yield two genetically identical gametic pronuclei, a stationary pronucleus and a migratory pronucleus, the remaining three haploid nuclei degenerate and disappear. Typically, they migrate to the posterior end of the cell where they remain as residual bodies until they disappear. In the present study we asked whether degenerating haploid nuclei share any properties with apoptotic nuclei. Specifically, we wondered whether they would be stained by "apofluor", a combination of vital fluorescent indicators that differentially stains apoptotic nuclei in living cells. "Apofluor" includes acridine orange, which becomes trapped in acidic compartments and stains lysosomal bodies a brilliant orange-red, and Hoechst 33342, which binds to DNA and stains nuclei bright blue. With this dye combination, while ordinary nuclei stain blue, the apoptotic macronucleus stains first blue-green, then yellow, and finally orange. The progression in color is presumed to be due to the accumulation of protons in the apoptotic nucleus compartment. We found that three of the four post-meiotic haploid nuclei, those that are eliminated, were stained differentially green, then yellow, and then come to be indistinguishable from the orange lysosomal bodies. Differential staining can occur even while the nuclei are located at the anterior ends of the cells, and before the "viable" nucleus divides to form pronuclei. These results indicate that haploid nuclei in the process of degradation are differentially stained in living cells by "apofluor", and that the differential staining occurs early in the elimination process. Further, since the degenerating haploid nuclei are stained by "apofluor" it is likely that they are degraded by a mechanism similar to the elimination of the apoptotic macronucleus.     

Phospholipids in plant and animal chromatin

Isolated hepatic nuclei and hepatic chromatin have been analysed for their DNA, RNA, protein and phospholipid content. The protein/DNA ratio is 3 for nuclei and 1.95 for chromatin extracted from Triton X-100 treated nuclei. The phospholipids, (2.36 +/- 0.91 (S.D.) per cent of the total nuclear material), are lost during the chromatin preparation mainly during the Triton X-100 washings of the nuclei. Nevertheless, 10 per cent of the total nuclear phospholipids remain bound to the chromatin. The comparative analysis of both nuclei and chromatin shows a difference in phospholipids and fatty acid composition. Thus, the chromatin-associated phospholipid cannot be attributed simply to contaminating nuclear membrane. This is supported by the autoradiographic study of semi-thin sections of interphase nuclei from root apices of Vicia faba in which [3H] ethanolamine is clearly localized in the chromatin and nucleolar regions of the nuclei.     

RNA synthesis in maturing avian erythrocyte nuclei

The rate of RNA synthesis and its inhibition by α-amanitin in the nuclei of mature and immature avian erythrocytes are increased with the increase in ionic strength of incubation medium. Polyacrylamide gel electrophoresis indicates that heterogeneous species of RNAs are synthesised in the mature and immature erythrocyte nuclei. However, a large number of high molecular weight RNAs are synthesised in the nuclei of immature erythrocytes. Elution profiles on poly(U)-sepharose chromatography indicate that the RNAs synthesised in the nuclei of two types of cell contain poly(A) segments. Sixteen per cent of mature erythrocyte nuclear RNA syntbesised are polyadenylated, while it is 13% in immature erythrocyte nuclei. However, the total RNA synthesised is 2–3 fold higher in immature erythrocyte nuclei than that in mature erythrocyte nuclei.     

Microspectrophotometry of cell nuclei stained with the Feulgen reaction. IV. Formation of tetraploid nuclei in rat liver cells during postnatal growth

1. DNA contents of the individual parenchymal nuclei of rat livers during postnatal growth were estimated by microspectrophotometric apparatus, and different ploidy classes of nuclei were classified by their DNA contents. With the same material the total number of parenchymal nuclei in the liver was counted microscopically. 2. If the DNA content of nuclei encountered most frequently in several tissues represents the diploid class, the ploidy classes of the rat liver cell nuclei correspond to di-, tri-, tetra-, and octoploid, with the di- and tetraploid ones predominating considerably. 3. In suckling rats (below 25 gm. of body weight) the liver parenchyma is composed almost exclusively of cells with diploid nuclei, whereas in young rats (above 80 gm.), of tetraploid nuclei. In the growth stage between 25 and 80 gm., there is a remarkable replacement of the diploid nuclei by the tetraploid ones. However, in the liver of adult rats weighing more than 150 gm., any increase or decrease in the frequency of diploid and tetraploid nuclei is hardly observable. In such rats, the nuclear population of the liver parenchyma seems to reach a cell-ecological equilibrium which is considered to be a stable one. 4. It is shown that such nuclear populations and the total number of nuclei in a liver are controlled by the growth state, and not by the age. 5. The decrease in the total number of diploid nuclei and the increase in tetraploid nuclei in the growing livers of rats weighing from 40 up to 130 gm. can both be explained by the hypothesis that the tetraploid nuclei originate from the interphase diploid nuclei without involving mitosis. This hypothesis implies that mitosis is confined to the reproduction of diploid cells alone. 6. It is suggested that, in general, the synthesis of DNA does not necessarily result in the formation of visible mitotic chromosomes. 7. Mitotic time and generation time of diploid nuclei and the percentage of the tetraploidization from diploid nuclei are calculated and discussed.     

STUDIES ON RNA SYNTHESIS IN TWO POPULATIONS OF NUCLEI FROM THE MAMMALIAN CEREBRAL CORTEX

A method is presented for the rapid separation of cell nuclei from the rabbit cerebral cortex into two populations. The first of these consists largely of nuclei with the morphological characteristics of neuronal nuclei, the second almost entirely of nuclei with the morphological characteristics of glial cell nuclei. From studies based upon sensitivity to the toxin α-amanitin, the ratio of incorporation of different bases, ionic requirements and differential sensitivity to actinomycin D, it is concluded that under both the classical low and high salt conditions described by other workers, two enzymes are active in RNA synthesis. The presence of a third enzyme of low activity cannot be excluded. No qualitative difference in the number of enzymes involved in RNA synthesis in neuronal and glial cell nuclei has been found, but there are quantitative differences in activity between the two nuclear populations.     

ISOLATION OF NUCLEI FROM YEAST

A method for isolation of nuclei from Saccharomyces cervisiae in high yield is described. The DNA/protein ratio of the isolated nuclei is 10 times higher than that of whole cells. Examination of these nuclei in phase and electron microscopes has shown them to be round bodies having a double membrane, microtubules, and a dark crescent at one end. The optimum conditions for extraction and resolution of histones of these nuclei on acrylamide gels have been investigated. The nuclei have an active RNA polymerase (E.C. 2.7.7.6) and are able to synthesize RNA in vitro. They are also readily stainable with Giemsa's, Feulgen's, and acridine orange methods.     

Numbers of muscle nuclei and myosatellite cell nuclei in red and white axial muscle during growth of the carp (Cyprinus carpio)

We determined the percentages of muscle fibie nuclei and satellite nuclei over a growth range of carp ( Cyprinus carpio ), as the increase in the number of muscle fibre nuclei is an important aspect of the increase in muscle mass, and myosatellite cells are believed to be the source of new muscle fibre nuclei. In white as well as in red axial muscle the percentage of the nuclei present in muscle that are muscle nuclei (muscle fibre nuclei+myosatellite nuclei) remained constant during growth (54 and 32% respectively). The difference in the percentage of non-muscle nuclei between white and red axial muscle is mainly caused by the higher content of endothelial nuclei in red axial muscle.
In white axial muscle the DNA/protein ratio (nucleus/sarcoplasm ratio) decreased between 3 and 15 cm S.l. In red axial muscle we found a continuous decrease in DNA/protein ratio over the entire investigated size range (3–50 cm s.l.). This may be related to a longer occurrence of hyperplasia in red than in white axial muscle.
In both fibre types the percentage of muscle nuclei being myosatellite nuclei decreased with increasing length, In white axial muscle it decreased from about 5% in carp of 5 cm s.l. to less than 1% in carp of 20 cm S.L.; for red muscle these values were 11 and 3% respectively.
For white axial muscle we calculated that, especially in larger fish, the myosatellite ceils alone cannot account for the increase in the number of muscle fibre nuclei during growth. The percentage of proliferating nuclei in muscle tissue, measured by the uptake of 5-bromo-2'-deoxy-uridine, is high enough to account for the total increase in nuclei. So indirect evidence is available that another cell type present in the muscle tissue may also be involved in the formation of additional muscle fibre nuclei.     

The simultaneous quantification of dopamine,norepinephrine and epinephrine in micropunched rat brain nuclei by on-line trace enrichment HPLC with electrochemical detection: Distribution of catecholamines in the limbic system

A method is described for the simultaneous quantification of dopamine, norepinephrine and epinephrine in microdissected rat brain nuclei using on-line trace enrichment high performance liquid chromatography (HPLC) and electrochemical detection (EC). The method is specific, sensitive and rapid with lower limits of detection comparable to those of radioenzymatic and mass fragmentographic techniques. The ideal application of the method to determinations in discrete microdissected brain nuclei is illustrated by mapping the topographic distribution of these three catecholamines in the limbic system of the rat. Dopamine is well represented and remarkably heterogeneously distributed in the amygdaloid nuclei. Norepinephrine concentrations are high in the interstitial (bed) nucleus of the stria terminalis and the paraventricular, periventricular, dorsomedial and medial preoptic hypothalamic nuclei, though generally more evenly distributed among the limbic nuclei than dopamine. Epinephrine was reliably quantified in a minority of the limbic nuclei examined with the highest concentration found in the paraventricular hypothalamic nuclei. The method described is ideally suited for examination of the pharmacological and functional heterogeneity of dopamine-, norepinephrine- and epinephrine-containing neuronal systems at the level of discrete brain nuclei.     

Differential accumulation of oocyte nuclear proteins by embryonic nuclei of Xenopus

Oocyte nuclear proteins of Xenopus are distributed into the cytoplasm of the maturing egg after germinal vesicle breakdown. Later they are found in all cell nuclei of the embryo. At early stages of development, different nuclear proteins behave differently. A class of 'early shifting' antigens is accumulated by pronuclei and cleavage nuclei, whereas others appear to be excluded from the nuclei at early stages but are shifted into the nuclei at blastula or during and after gastrulation. Accumulation of 'late-shifting' nuclear antigens is a gradual process and occurs during a period characteristic of each protein. Multiple artificial pronuclei can be formed after injection of sperm nuclei, erythrocyte nuclei or pure lambda-DNA into unfertilized eggs. The artificial pronuclei accumulate early- but not late-shifting proteins. Early-migrating proteins rapidly accumulate into the germinal vesicle after de novo synthesis in the oocyte, indicating that the efficiency of translocation into nuclei is an intrinsic property of each protein. Artificial extension of the length of the cell cycle before midblastula transition does not lead to accumulation of the late-shifting nuclear antigens investigated.