Breakthrough in Image Analysis Software for Cancer Research
Mark Edward Plaskow ; Nationwide Childrens Hospital;
Content:
The Research Informatics Core Image Analysis software enables automated electronic scanning and analysis of digital pathology images to locate markers for both pediatric and adult cancerous diseases and offer enhanced and time-saving tools for pathologists and reviewers. Similarities and differences between cancerous diseases were analyzed, creating an ability to search across specimens from many protocols representing many different diseases.
Technology:
The Image Analyzer utilizes the .NET technology, in addition to integration with virtual microscopy software, and a Microsoft SQLServer 2005 database for efficient location, persistence, indexing, and retrieval of key areas of interests on analyzed slide images. It runs as both a web-enabled application and a windows compiled application.
Design:
The image analysis application was designed from a use case developed within the Childrens Oncology Group (COG). The use case included defining a pediatric cancer protocol and scanning all cases in the protocol with associated slides. Consultation was conducted with solid tumor (sarcoma) specialists and hematology and leukemia experts.
Results:
The first release of the Research Informatics Core teams Image Analysis software was dedicated to electronic location of abnormal nuclei diffusion within cancerous tumors. An advanced algorithm gives case pathologists and reviewers areas of interest when used against an individual slide specimen image, subset, or entire image archive. The Reviewer can inspect the returned data and corresponding digital images using a virtual microscope/image viewer, locating, zooming/adjusting, and annotate desired parts of the image. Subsequent releases aim at a true pathologist dashboard, with the ability to specify criteria across cancerous diseases, including both solid and liquid infections, searching equally on nuclei or white blood cell clustering, cellular mytosis, infectious cellular growth, etc. Solid tumor specimens are searchable by tumor necrosis levels, while cancerous blood infections are searchable by white blood cell developmental stages, platelet and red blood cell levels, etc. The tool can search for custom cells which are definitive markers (i.e. Auer Rods in Acute Myelogenous Leukemia).
Conclusion:
In conclusion, our Image Analysis is doing a great deal of good within cancer research, providing rapidly available tools for Pathologists in the identification of specific case types and cancer characteristics using virtual microscopy.
