Yale's Center for Medical Informatics Receives $1.5 Million Grant To Test Next Generation Internet
Yale researchers have received a $1.5 million grant to develop a computer system that will analyze cancer cells and compare them against thousands of other specimens.
The grant awarded to the Center for Medical Informatics will be used to analyze two types of cancer cells – lymph and thyroid – using a high performance computer network.
The goal is to use the PathMaster computer system to help pathologists in making their diagnoses, and to test the capabilities of the Next Generation Internet (NGI) in a research program supported by the National Library of Medicine.
“When someone sends in new images from a specimen each will be analyzed and compared against this database to see which cells match best,” said Perry Miller, M.D., professor of anesthesiology at Yale School of Medicine and director of Medical Informatics. PathMaster was developed by Mark Mattie, M.D., an associate research scientist in pathology who works closely with Miller.
The lymphoma cells and cells aspirated from thyroids that make up the database will be obtained primarily at Yale-New Haven Medical Center. Images of the top-ranking specimens from the database then will be sent back to the pathologist to help with the doctor’s diagnosis.
Miller said the lymphoma and thyroid cells are relatively easy to obtain and lend themselves to examination as single cells rather than as tissue samples. The database currently contains about 500 images. The goal is to reach 30,000 images over the next three years.
“PathMaster is designed to help the pathologist in the process of making a diagnosis based on a set of cells seen in a cytologic specimen,” he said. “The project will focus on helping with the diagnosis of a single cell, or set of single cells, for which the pathologist would like diagnostic assistance.”
“We believe that PathMaster’s approach will ultimately prove valuable for a wide range of cytology domains,” Miller added.
In addition to providing a tool for pathologists, he said the PathMaster program is one of several projects approved by the National Library of Medicine to test the capabilities of the NGI.
“The nation is interested in building the NGI and the National Library of Medicine wants to help define what the needs of medicine will be,” Miller said. “The Internet as it is today is slow and has security limitations. The NGI will be much faster with a much higher bandwidth, which is needed for remote use. Images are a good test of the NGI. The real purpose of this grant is to develop PathMaster specifically as a way to test the network characteristics it will require.”
NGI capabilities and characteristics of interest to the National Library of Medicine include quality of service, medical data privacy and security, nomadic computing, network management, and infrastructure technology for scientific collaboration. The PathMaster project will focus on quality of service and network management.
The digitized cell images incorporated into PathMaster will be automatically analyzed to determine more than 2,000 mathematically derived descriptors. To speed up the analysis, the computation will be broken down into a variety of subcomputations. The subcomputations will be performed in parallel by computers linked initially by a high performance local network, using the Linda parallel computing language developed by Yale Computer Science Professor David Gelernter.