Find Funding

Contract Topics

234 Develop Automated Methods to Identify Environmental Exposure Patterns in Satellite Imagery Data

Number of anticipated awards: 1-2
(Fast-Track proposals will not be accepted.)
Budget (total costs): Phase I: $150,000;
Phase II: $750,000

The deadline for receipt of all contract proposals submitted in response to this solicitation was:
5:00 p.m. Eastern Standard Time
Monday, November 6, 2006

The NCI seeks to develop an automated process to apply existing pattern recognition algorithms to satellite image data and present the results to the researcher that employs a user-friendly, interactive geovisualization tool. Useful features of the automated process include options to use smoothing techniques, cluster identification, change detection, isopleth map production and animation to clearly illustrate the pattern-identified images of health- (preferably cancer-) related environmental measures over time. It would also be important to include, if possible, any existing algorithms to convert raw image values to categorized factors, such as crop maps, soil content or water quality to permit the geovisualization tool to operate on these derived, as well as original, values.

A requirement of the process will be that it can manage, process, and distribute satellite images, and orthorectify (geographically register) the piecewise images from multiple satellite passes over a geographic area and present a unified image of the entire area. This is an important first step to be able to use satellite imagery for environmental exposure assessment on a routine basis.

Location-specific data have become increasingly important for cancer research. Cancer mortality, morbidity, and survival rates are known to vary by place. A growing body of research has shown the importance of local neighborhood influences on cancer outcomes, risk behaviors, and access to care (Kirby and Kaneda 2005; Cubbin 2000). Researchers in cancer surveillance, cancer control, health disparities, and other areas now routinely collect information on location of patient residence and potential risk factor exposures from a variety of sources.

GIS software capabilities and statistical methods for analyzing geospatial data have advanced rapidly over the past decade. When NCI first published an atlas of cancer mortality in 1975, only the National Oceanic and Atmospheric Administration had software locally available to produce the maps. Now, maps can be produced on a desktop PC or handheld PDA using available software. The volume of data used for location identification has increased rapidly, due to technological advances through a Global Positioning System (GPS) that has input data from the ground or from satellite imagery. These high volume data streams or images must be processed quickly and have the ability to integrate into existing geospatial databases.

Satellite imagery has been used to estimate an individual's probable exposure to agricultural pesticides (Ward 2000; Nuckols 2004) and water quality (Xiao 2006). LandSat satellite images are now available for many areas from the past 30 years, making them a potentially useful source for identification of historic environmental exposures for cancer studies. While methods exist for detecting patterns and converting image data to potential exposure estimates, the volume of satellite data has outgrown the application potential for these methods to effectively estimate exposures. The integration of millions of available time-specific images is prohibitive.

Currently, it is not possible to integrate images or data streams from numerous data sources to attempt smoothing techniques, cluster identification, isopleth map production and animation to analyze the pattern-identified images over time when multiple years of cancer and potential exposure data are under consideration. Therefore, inferences between cancer outcomes and potential exposures may not be realized.

Cancer research and health research have not grown technologically at the same pace. Frequently cancer and health researchers are unable to take advantage of the technological advances seen in other scientific fields due to database and methodological limitations.

Phase I Activities and expected deliverables:

  • Review the literature to determine types of environmental exposures that have both been associated with increased cancer risk and can be measured from satellite data. Determine what, if any, algorithms exist to convert raw satellite image values to exposure measures.
  • Determine what, if any, commercial software products exist that may serve as a platform for the proposed geovisualization tool. For example, the proposed product might be best developed as an add-on to an existing GIS, image analysis or data visualization package.
  • Evaluate the availability of satellite images over the U.S. by time for a representative set of desired exposure measures. Identify problems with comparability and availability of data over time and as collected by different satellite monitoring systems.
  • Convene a focus group of environmental epidemiologists and other interested scientists to solicit input on the functionality required for the proposed product.
  • Develop a statement of functional requirements and user interface requirements for the product.
  • Develop a working prototype of the system using a sample of existing satellite imagery data.
  • Include funds to present Phase I findings and demonstrate product prototype to NCI staff.

Phase II Activities and expected deliverables:

  • Conduct a formal usability study of the software with representative users to evaluate the prototype developed in Phase I. Enhance and modify the prototype's functionality and user interface based on this feedback.
  • Demonstrate the accuracy of the derived patterns in the images by comparing results of application of the software to published results or by simulating images that appear similar to that used in published work.
  • Demonstrate the flexibility of design that would permit updating the software as new image categorization algorithms or satellite image formats are added.
  • Complete the development of the full software package, including technical documentation.
  • Identify Phase II barriers to evaluating the impact of the software and resolutions to these barriers.
  • In the first six months of the first year of the contract, provide the program and contract officers with a letter of commercial interest.
  • In the first six months of the second year of the contract, provide the program and contract officers with a letter of commercial commitment based on the successful outcome of the Phase II.
  • Include $24,000 in the budget for an independent vendor to evaluate the final product.
  • Include sufficient travel funds for the P.I. to participate in an NCI/DCCPS SBIR Showcase.
  • Prepare at least one manuscript describing the development and evaluation of the product for publication in a peer-reviewed scientific journal.
  • Submit final report in the template provided by the NCI program officer.

[Back to Contract Topics]