Patch-Based Nonlinear Image Registration for Gigapixel Whole Slide Images.

OBJECTIVE: Image registration of whole slide histology images allows the fusion of fine-grained information-like different immunohistochemical stains-from neighboring tissue slides. Traditionally, pathologists fuse this information by looking subsequently at one slide at a time. If the slides are digitized and accurately aligned at cell level, automatic analysis can be used to ease the pathologist's work. However, the size of those images exceeds the memory capacity of regular computers. METHODS: We address the challenge to combine a global motion model that takes the physical cutting process of the tissue into account with image data that is not simultaneously globally available. Typical approaches either reduce the amount of data to be processed or partition the data into smaller chunks to be processed separately. Our novel method first registers the complete images on a low resolution with a nonlinear deformation model and later refines this result on patches by using a second nonlinear registration on each patch. Finally, the deformations computed on all patches are combined by interpolation to form one globally smooth nonlinear deformation. The NGF distance measure is used to handle multistain images. RESULTS: The method is applied to ten whole slide image pairs of human lung cancer data. The alignment of 85 corresponding structures is measured by comparing manual segmentations from neighboring slides. Their offset improves significantly, by at least 15%, compared to the low-resolution nonlinear registration. CONCLUSION/SIGNIFICANCE: The proposed method significantly improves the accuracy of multistain registration which allows us to compare different antibodies at cell level.

Identification of genes involved in taura syndrome virus resistance in litopenaeus vannamei.

Abstract The goal of the present research was to identify the genes that are differentially expressed between two lineages of Pacific white shrimp Litopenaeus vannamei displaying different susceptibilities to Taura syndrome virus (TSV) and to understand the molecular pathways involved in resistance to the disease. An oligonucleotide microarray was constructed and used to identify several genes that were differentially expressed in the two L. vannamei lineages following infection with TSV. Individual L. vannamei from either resistant or susceptible lineages were exposed via injection to TSV. Individuals were removed at 6 and 24 h postinfection, and gene expression was assessed with the in-house microarray. The microarray data resulted in the selection of a set of 397 genes that were altered by TSV exposure between the different lineages. Significantly differentially expressed genes were subjected to hierarchical clustering and revealed a lineage-dependent clustering at 24 h postinoculation, but not at 6 h postinoculation. Discriminant analysis resulted in the identification of a set of 11 genes that were able to correctly classify Pacific white shrimp as resistant or susceptible based on gene expression data. Received June 21, 2013; accepted October 24, 2013.

Epidemiological parameters of White Spot Syndrome Virus infections in Litopenaeus vannamei and L. setiferus.

An experimental protocol based on a mathematical epidemiology model was developed to study the transmission, virulence, and recovery rates of White Spot Syndrome Virus (WSSV). Two modes of transmission were compared for WSSV in Litopenaeus vannamei. We compared transmission by ingestion of infected cadavers to transmission by cohabitation with infected animals. In addition, we compared the ingestion transmission of WSSV in L. vannamei and in L. setiferus. Finally, we compared the virulence and recovery rates of WSSV in L. vannamei and L. setiferus. The transmission rate of WSSV to L. vannamei by cohabitation was 0.01. The transmission rate by ingestion of infected cadavers was over an order of magnitude larger at 0.46, suggesting that cohabitation is a much less important mode of transmission for WSSV. A statistically significant difference was detected between the estimates of ingestion transmission of L. vannamei (0.46) and those of L. setiferus (0.84), yet no differences in the virulence or recovery rates were detected between hosts. The overall estimated virulence rate was 0.34, and the overall estimated recovery rate from a WSSV infection was 0.007 for both species. According to epidemiological theory the threshold density of hosts necessary for an epidemic to occur is directly related to the virulence and recovery rates and inversely related to the transmission rate. Therefore, the epidemic threshold density may be lower for ingestion transmission than cohabitation transmission and lower for L. setiferus than for L. vannamei.

Transmission of white spot syndrome virus (WSSV) to Litopenaeus vannamei from infected cephalothorax, abdomen, or whole shrimp cadaver.

Shrimp viruses can remain infectious in frozen shrimp tissue and have been found in frozen commodity shrimp. Therefore, the threat of viral outbreaks in wild and cultured shrimp via frozen commodity shrimp exists. Because frozen shrimp are imported with and without the cephalothorax, more knowledge is needed concerning the infectivity of a cephalothorax relative to that of an abdomen. We compared the mortality rates from shrimp exposed to a WSSV-infected cephalothorax, abdomen, or whole shrimp cadaver. Estimates of transmission coefficients from the exposures to the infected cephalothorax, abdomen, or whole shrimp were also calculated because the transmission coefficients account for differences in the initial doses. In addition, we compared the variability in infectivity of pieces of shrimp by feeding 24 equal-sized pieces of cephalothorax and abdomen to 24 individually isolated shrimp. In Expt 1, susceptible shrimp did not completely consume the infected abdomen, and a significant difference was detected among shrimp exposed to the abdomen (mortality rate = 0.40), cephalothorax (mortality rate = 0.75), and whole shrimp cadaver (mortality rate = 0.67). The calculated transmission coefficients were 0.95 from an infected cephalothorax, 0.59 from an infected abdomen, and 0.69 from an infected whole shrimp cadaver. In Expt 2, susceptible shrimp were starved to ensure complete ingestion of each dose. No significant difference was observed in the estimated mortality rates from an infected cephalothorax (0.58), abdomen (0.63), or whole shrimp (0.67). The calculated transmission coefficients were 0.84 from an infected cephalothorax, 0.83 from an infected abdomen, and 0.60 from an infected whole shrimp cadaver. In Expt 3, no difference was observed in the mortality rates resulting from exposures to pieces of infected cephalothorax (0.57) or abdomen (0.58). Our results suggested that there was no difference in the viral loads of a WSSV-infected cephalothorax or abdomen, but that the cephalothorax was more infectious, probably because it was more palatable. In addition, our results are inconsistent with some assumptions of pathogen transmission used in epidemiological models. Some shrimp may be less aggressive feeders; therefore, susceptible shrimp are differentially contacting the dead infected shrimp in the exposure tanks, violating the random mixing assumption. Moreover, virus is probably not homogeneously distributed throughout an infected shrimp, suggesting that contacts between susceptible and infected shrimp are not equally likely to result in transmission.

Parasitology meets ecology on its own terms: Margolis et al. revisited.

We consider 27 population and community terms used frequently by parasitologists when describing the ecology of parasites. We provide suggestions for various terms in an attempt to foster consistent use and to make terms used in parasite ecology easier to interpret for those who study free-living organisms. We suggest strongly that authors, whether they agree or disagree with us, provide complete and unambiguous definitions for all parameters of their studies.

Quantitative aspects of evaluating the consequences of pollution for parasite populations and communities.

The utility of parasites as indicators of environmental health depends upon scientists' ability to understand how environmental stress may cause changes in parasite populations and communities. Predicting the changes due to pollution calls for dynamic models and experimentally determined knowledge of the consequences of pollution on the parameters of the model. Through an interplay of modeling, laboratory experiments, and field comparisons scientists can develop a theory of pollution and parasites and perhaps use it to assess the impact of pollution. To develop a theoretical framework for the effects of pollution on host-parasite relationships will require an identification of the appropriate scale and a determination of whether the host and parasite populations should be treated as open and recruitment-driven or closed and reproduction-driven. A theory can then be developed using epidemiology models. Those models are divided into prevalence models that track the prevalence of infection and abundance models that track the number of parasites per host. Whether a deterministic model or a stochastic counterpart should be used depends upon the scale but generally a stochastic model will give a more realistic picture. Mimicking the variability inherent in natural systems is the goal of stochastic models. The models emphasize that field tests of their predictions can only be made by studies designed with extensive replication of samples from reference and stressed sites.

Recruitment-driven, spatially discontinuous communities: a null model for transferred patterns in target communities of intestinal helminths.

Populations and therefore communities of intestinal helminths of vertebrates are fueled by recruitment of new individuals from outside the host. The source of new individuals is often an intermediate host that harbors several infective propagules of 1 or more species. Hence these source communities are transmitted in packets of infective propagules to target communities in definitive hosts. Packets not only provide recruits to target communities, but, because a packet of propagules possesses its own structure, it may also transmit structure to the target community. We use this system to examine the contribution that structure in the source pool of propagules makes to the structure of recruitment-driven target populations and communities. By treating the dynamics of such target populations and communities as immigration-death processes, we conclude: (1) Unlike a birth-driven population a recruitment-driven target population will grow to an asymptotic limit even in the absence of density-dependent processes or reaching carrying capacity; (2) the frequency distribution of the number of recruits entering target populations will determine the frequency distribution of adults in target populations; (3) interspecific associations among species in the source community will be transmitted to target communities, but the magnitude of the transmitted associations will depend upon the relative survival rates of the species; and (4) for associations of equal magnitude in a source community, the magnitude of a transferred negative association will be less than the magnitude of a positive association in a target community. Two examples of source communities in salt marsh crabs reveal that source infracommunities exist with the hypothesized structure. Further, the source helminth communities display a greater number of positive than negative interspecific associations. The inequity in transfer and the existence of a greater proportion of positive associations in source communities may explain the widespread occurrence of excess positive associations that has been noted in recruitment-driven communities.

Excess positive associations in communities of intestinal helminths of bats: a refined null hypothesis and a test of the facilitation hypothesis.

The null hypothesis that the number of positive pairwise covariances should equal the number of negative pairwise covariances in samples from communities of randomly associated helminth species was reevaluated. The proportion of positive covariances in a sample from a community of independent species depends upon the proportion of rare species (prevalence less than 10%), the proportion of common species (prevalence greater than 90%), and the size of the sample of hosts. If rare species dominate, then there will be an excess of negative associations; if common species dominate there will be an excess of positive associations. Many helminth communities have more rare than common species, therefore samples from communities that show an equal number of positive and negative covariances have a greater number of positive associations than is expected for randomly associated species. Increased sample size will reduce the sampling bias, but at least 100 hosts are necessary and often 500-7,500 hosts are required. The excess of positive covariances between helminth species in 10 populations of bats disappeared after restricting the analyses to hosts in which both members of a species pair were present. This result suggests that excess positive associations between helminth species in bats are due to joint presences and absences in hosts rather than to interspecific facilitation. Interspecific facilitation would be supported by observed positive correlations between the intensities of individuals of the species pairs.

The role of positive and negative interspecific associations in the organization of communities of intestinal helminths of bats.

Twelve populations of bats were examined to determine the extent of interspecific associations in determining the species richness of intestinal helminth infracommunities. The pool of helminth species which was available to individual bats ranged from 2 to 21. The 'summed binomial' distribution was determined to underlie the host frequency distribution of the number of helminth species per host. Overall covariation in occurrences of species in replicated communities can be detected by testing for the equality of the observed variance of the host frequency distribution to the variance expected when species are allocated to hosts at random. Where statistically significant the covariance was indicative of a majority of positive rather than negative interspecific associations. As the mean number of species per host in a host population increases not only does the number of positive associations increase but so does the proportion of species pairs which exhibit positive associations. Although there is an increase in the proportion of species pairs which exhibit positive associations as the number of species increases, the magnitude of the associations (as indicated by the mean positive or the mean negative pairwise covariances) does not. Therefore, we concluded that positive interactions are more common than negative interactions in determining the species richness of helminth infracommunities of bats. Further, positive associations become even more important as the community becomes more complex. However, the increased importance is derived from the number rather than the strength of the associations.