Description: In collaboration with Professor Pietro Perona of Caltech's Electrical Engineering department, we have initiated development of a neural network-based image processing system to automate pollen counting and identification. Based upon initial results with laboratory generated samples and measurements of particles collected with the Burkard sampler, we have modified the Burkard sampler to reduce the particle density so that individual particles can be better identified. We now seek to train the automated pollen counter on samples collected using that improved sampler. To do so requires generation of at least 1000 micrographs containing identified pollen grains of each species of pollen. Although we have achieved that with several of the most abundant species, manual collection of such large numbers of pollen images for the 18 pollen species that we are initially classifying remains a formidable task. Ultimately, we require a library of 1000 images of each of the 30 pollen types, and 16 fungal types, collected from the Pasadena air. We plan to complete this library within 12 months. The Southern California Health Sciences Center Pilot Project Award will facilitate collection of the large numbers of pollen images needed to train and validate the performance of the automated pollen counting system by enabling purchase of a computer-controlled microscope that will enable fast, efficient processing of air samples and alleviate the need for laborious manual microscopy and image capture. Our initial focus with this system will be on providing the pollen grain identification that is required to train the pollen analyzer. A microscope with computer-driven stage and auto-focus will enable work to be undertaken under this study will include:
•  Integration of the computer controlled microscope with the pollen identification software to facilitate automated scanning of Mylar tape samples collected using the Burkard sampler to locate and capture images of pollen grains.
•  Use of that system to capture images of hundreds to thousands of individual pollen grains of the major pollen species in the Pasadena atmosphere.
•  Expert identification of the pollen grains captured in those images to create a library of identified pollen particles.
•  Use of that library to train the pollen counting and identification software.
•  Validation of the trained system by analysis of independently collected samples.
The primary goal of this research is to complete development of a pollen counting and identification system that is sufficiently robust and efficient to facilitate collection of ambient pollen data at larger numbers of sites than is presently practical. Moreover, automation will reduce biases and poor statistics in pollen counting that result from operator fatigue with present manual counting. We anticipate being able to increase the numbers of particles of each species counted, and to undertake investigations of temporal trends in pollen concentrations that are not possible with present manual data collection.

---

Description:

We hypothesize that exposure to ambient air pollution associates with atherogenesis. We use the carotid artery intima-media thickness (IMT) as the quantitative marker of sub-clinical atherosclerosis. Participants of two clinical trials conducted recently in Los Angeles provide the baseline IMT data to investigate the cross-sectional association between IMT and ambient air quality. Our primary indicator of urban air pollution will be particulate matter up to 2.5 µm in diameter, PM 2.5 . We will use a PM 2.5 surface that has recently been developed for the South California Los Angeles area. PM 2.5 concentrations will be assigned to the street address of both the residential and the work place location of each subject. Ambient ozone, NO 2 , and CO concentrations, derived from land use regression models will be assigned to each subject. Questionnaire based information regarding commuting patterns will be used to refine exposure estimates. Regression models will estimate the association between the assigned exposures and IMT. The associations will be tested in subgroups, determined by sex, age, and other potential modifiers of effects. The pilot study will produce important results, including: (1) initial insights into the potential contribution of air pollution to atherosclerosis; (2) contributions to other applications using other populations as well as longitudinal data from USC trials; and (3) enhance the design optimizations of a pending EPA STAR grant, if awarded later this year.

---

Description: The role of dietary fat in the development of colorectal cancer is inconclusive, particularly for monounsaturated and polyunsaturated fatty acids. Animal and in vitro studies support a role of polyunsaturated fatty acids (PUFAs) in colon carcinogenesis. In particular, high intake of w -6 PUFAs and low intake of w -3 PUFAs are linked to increased colon cancer risk; therefore, the dietary ratio of w -6: w -3 PUFAs seems to be critical, given that w -3 PUFAs may inhibit the action of w -6 fatty acids. Dietary w -6-PUFAs serve as substrates for cell membrane w -6-PUFAs, which can be released into the cytosol and serve as precursors of eicosanoid synthesis. Reactive oxygen species (ROS) generated during inflammation, eicosanoid biosynthesis, smoking, among others, can oxidate cell membrane w -6-PUFAs generating lipid hydroperoxides, which can either be reduced by glutathione peroxidases or can become reactive compounds able to attack DNA and induce mutations. We hypothesize that polymorphisms in w -6-PUFA metabolism and DNA repair genes may modify the effect of dietary unsaturated fat in colorectal cancer. If this hypothesis was true, taking into account the genotype of proteins that participate in the PUFA-DNA damage/repair pathway may help better understand the role of unsaturated fatty acids and colorectal cancer risk. We propose to test our hypotheses obtaining two types of DNA repair variation measures in colorectal cancer cases and controls. One, a direct measure of DNA repair proficiency using the single cell electrophoresis assay (Comet assay). Two, by studying polymorphisms in genes that play critical roles in the metabolism of w -6-PUFAs and repair of w -6-PUFA-induced damage. We will analyze our data using conventional logistic regression-based methods, as well as a more innovative approach using Bayesian modeling averaging of the PUFA-DNA damage pathway. We will conduct this study within the USC component of the Colorectal Family Registry (CFR, 743 cases-controls), an international consortium funded by NCI and formed by six research centers in the US , Australia , and Canada . This registry is the largest in its kind (7,000 patients) and consists of colorectal cancer patients and their families, who provide demographic and nutritional data, biological specimens, and family history of cancer information. Our long-term goal is to test our hypotheses using the entire CFR, in collaboration with investigators in the other centers. The results we will obtain with this proposed pilot project will provide the needed preliminary evidence to seek R01 funding to carry out a CFR-wide project.

---

Description: It has been suggested that early exposure to infections may influence the Thl/Th2 cytokine balance in infancy, resulting in a skewed immune response which would in turn, increase susceptibility to asthma and other diseases later in life. One report suggests that the mechanism for early regulation ofThl/Th2 cytokine balance may be methylation of the interferon-y (LNF-y) promoter region, possibly in response to infection (White, 2002). We are conducting a pilot study with funds from SCEHSC to develop a method to detect ThI/Th2 cytokines in small blood samples in order to evaluate cytokine levels from an infant heel stick. Using funds from last year, we have already learned with this pilot study that we can detect cytokines in as little as 250 µl of whole blood by CBA and that these measurements are highly correlated to using ELISA. Although we have shown this in replicates from one blood sample, we need validate the test on a larger scale. In addition, we will collaborate with a pediatrician in Watsonville to obtain sequential blood samples in infants beginning with cord blood through 18 months old, to evaluate INF-y methylation at different ages and its effect on cytokine levels at early ages. The purpose of this pilot study is to produce preliminary data for a larger pregnancy cohort in which we will employ these methods to learn about immune system development from birth and test hypothesis that relate Thl/Th2 disposition to asthma later in life.

---

Description: Chronic sinusitis is a common health problem in the United States and is responsible for excess morbidity and consumption of medical services, however little is known about risk factors. We propose to conduct a hypothesis-generating pilot study to test the association between air pollution and a marker for chronic sinusitis by measuring sinus mucosal thickness on head and neck CT scans from subjects residing in very polluted areas compared to those residing in minimally polluted areas. Areas of high and low air pollution are defined using air monitoring stations set up by the NIEHS Exposure Assessment Core (Avol). Head and neck CT scans will be obtained from the major hospitals in each of the identified regions. Sinus mucosal thickness will be measured independently by two USC ENT fellows. The mucosal thickness will be measured in the maxillary and anterior ethmoid sinuses. These are the sinuses most likely to be involved in any inflammatory process. Both sinuses will be evaluated in the axial and coronal planes and the thickest mucosa recorded. Measurements will be made directly with a millimeter ruler. The Biostatistics Core will provide assistance with analysis comparing the proportion of persons with abnormally thickened sinuses (> 1 mm) to those with normal sinuses ( < 1 mm). Adjustment for race, gender and age will be included if we find an association with any of these demographic variables. We estimate that we can review 120 scans in each community over a 4-6 month period, as these hospitals average approximately 30 non-sinus related head and neck CT scans per month. This will conservatively give us 77% power to detect a relative risk of 2.3 and 84% power to detect a relative risk of 2.5. If we find any association between residence in a community with higher levels of air pollution and abnormal sinus mucosa thickness, we will plan an R01 submission to NIH to test the hypothesis in a formal study with individual data.

---