Healthcare agencies interested in improving patient care by sharing medical images in online archives must grapple with the challenge of creating efficient systems in these days of memory-hungry CT scans and MRIs. Read about the Wellcome Trust Sanger Institute's pilot database project sharing 40 terabytes of medical images using Oracle interMedia. interMedia is a part of Oracle Database that enables the efficient management and retrieval of image, video, and audio data. With Oracle Database 10g Release 2 with interMedia, Sanger's Atlas database holds metadata to help users store and search and view medical images, which adds value. The interMedia technology is also leveraged by Oracle Content Services 10g, part of Oracle Collaboration Suite 10g, to provide enterprise content management solutions for financial services, insurance, government, manufacturing, and healthcare customers.

As Published In February 2006

Learning to Share
By Jeff Erickson

The Sanger Institute prepares to dispense terabytes of biological images.

On an autumn morning in England, the staff of the Wellcome Trust Sanger Institute are removing their jackets and putting their personal posters back up after another royal visit to the institute's grounds. With a new wing of laboratories and IT facilities officially open and the ribbon cut by Her Royal Highness Princess Anne, most of the 800 researchers and technologists are getting back to work on the genomic discoveries for which the institute is famous. But a few of the researchers at Sanger are up to something else.

The Atlas of Protein Expression team will continue to work on producing and sharing 40 terabytes of biological images in an Oracle database. The database, known as Atlas, will allow researchers to peer deep into tissue samples. As a scientific effort, Atlas will help researchers learn where and when a gene does its work. As an IT project, it will provide an example of how healthcare organizations can build online healthcare archives in these days of memory-hungry MRIs and CT scans.

The Power to Share

"The Atlas has become a massive image processing project," says Martin Widlake, head of database services at the Sanger Institute and the Atlas team's database lead. In the Atlas, researchers aim to stain as many of the 24,000 proteins in mice as they can, and then take images at a resolution that will satisfy a pathologist or a histologist. "We have to make extremely high-resolution images accessible and searchable by research centers all over the globe," Widlake says.

That's a tall order, even for Sanger, the powerhouse institute that provided much of the heavy lifting for the global Human Genome Project. Established in 1993 and primarily funded by the US$20 billion Wellcome Trust, the Sanger Institute is accustomed to taking on big challenges. It was its successful completion of the genome for the nematode worm, in 1998, that proved the human genome could be sequenced. Sanger took a leading role in sequencing the human genome and has since sequenced several other genomes, including yeast, the mouse, and many bacteria. In the process of crunching and recrunching the millions of gene sequences for each project and creating the infrastructure to share its results with researchers worldwide, Sanger has become one of the premier IT organizations in Europe—and the pride of the U.K. "We maintain more than 350 terabytes of storage and one of the world's top computer facilities," says Widlake. "We also maintain a series of Oracle databases, including the main lab information management systems, which we use to track millions of samples through a complex scientific process."

Finding interMedia

After a process of discovery, Widlake's team decided that Atlas would also run on an Oracle database. "At first we were going to use a simpler database and pull the images from files, but we wanted Atlas to be a multiuser, interactive system, and files became too much work. Next we were looking at using Binary Large Objects (BLOBs), in which the database knows that it is storing a large collection of binary data but doesn't know that it's an image," Widlake says. "Then one day someone from Oracle's interMedia team called and said, 'You wouldn't have any need to handle large quantities of images in the database, would you?' We immediately changed our beta to include interMedia."

Oracle interMedia is a feature in Oracle Database that enables the efficient management and retrieval of image, audio, and video data. This same interMedia technology is also leveraged by Oracle Content Services 10g, part of Oracle Collaboration Suite 10g, to provide enterprise content management solutions for financial services, insurance, government, manufacturing, and healthcare customers. In Oracle Database 10g Release 2, Oracle interMedia introduced native support for DICOM (Digital Imaging and Communications in Medicine), the standard for storing and communicating medical images. DICOM is rich in metadata—as many as 2,000 attributes can be stored along with each image. Oracle has built a simple way to parse this metadata and use it in indexes and searches.

Oracle's interMedia team believes that simple and efficient storage, indexing, and retrieval of images across clinics, hospitals, and even regional medical organizations holds the potential for better and less-expensive healthcare. The benefit to researchers is more immediate.

"With interMedia, the database understands the fact that a piece of information is an image," says Widlake. In Oracle Database 10g Release 2, interMedia recognizes several formats, including JPEG, TIFF, DICOM, and JPEG2000. "JPEG2000 is important to us because it can compress data without losing information," says Widlake. "We're dealing with scientists, and they do not like to lose information."

Snapshot Wellcome Trust Sanger Institute www.sanger.ac.uk Funded primarily by the Wellcome Trust, the Sanger Institute is one of the leading genomics centers in the world. Location: Hinxton, Cambridge, United Kingdom Budget: US$430 million from 2001 to 2006 Employees: More than 800 Products and services: Oracle Database 10g, Oracle Real Application Clusters, Oracle interMedia, Oracle Recovery Manager

Search for Meaning

With genomes sequenced, Sanger's next step is to search them for meaning. "It's all well and good knowing that a particular gene translates into a particular protein," says Widlake. "But researchers need to know when the protein is being expressed, where in the tissue it's being expressed, and how strongly."

The Atlas team will attach a stain to an antibody of a particular protein in a mouse embryo and take slides of the stained tissue at various days in the embryo's development. "That is why the project is called an atlas," Widlake explains. "It provides a geography of where and when the gene is expressed in the mouse embryo."

The Atlas will display images of 1,000 x 1,000 pixels and greater as JPEG2000 files. Researchers worldwide will visit the gene Atlas with their browsers, enter the gene or protein they are interested in, and be able to view a detailed image of the gene or protein where it was found in the mouse embryo along with clinical information about the embryo.

Tackling Disease

"What we're doing is laying the foundation for preventing, treating, and curing disease by better understanding our biology," says Widlake, noting that except for the whiskers and the cheese fixation, mice are surprisingly similar to humans. "What happens in the mouse translates very well for a human," he adds. The endgame is to help researchers come up with preventive treatments and cures.

Widlake hopes the Atlas will help researchers develop genetic tests to know if a person is at risk for a certain cancer, or to understand how a person's genetic makeup causes them to react well or badly to a treatment that already exists. "The drug companies have thousands of drugs they know work," says Widlake, "but they can't use them because they have a serious negative effect on some people. But if they can understand why a particular drug harms some people and not others, they can track that down to a genetic difference, test for that difference, and begin using those treatments to help patients who don't have a genetic propensity for the negative effect."

The Atlas of Protein Expression team has completed a successful beta of their image-sharing database. "The interMedia team has been great at solving our issues," Widlake says. "We're confident in the technology."

How did you drive adoption of the new system?

For More Information LEARN more about Oracle Fusion Middleware standards-based products at oracle.com/products/middleware

The TUSC people did a very good job of teaching us how to use it. In fact, they recommended more training than we ended up doing, and in hindsight we wish we'd taken their advice. Some of the implementation phases would have been easier with another month of learning. It took a while for our people to understand the full scope of the changes.

How will this infrastructure help you in the future?

At our management meetings, we always ask ourselves if we are fully utilizing our technology, how we can strengthen and expand our services, how we can optimize communication. This open technology platform allows us to interface with other systems such as wireless communications and voice over IP. TUSC helped us build a solid foundation we can easily extend.

Q & A Alan Louie of Life Science Insights: Life-Saving Data Finds a Home How can technological and social hurdles be overcome to form regional and national healthcare archives? Alan Louie, Ph.D., research director of Life Science Insights, in Framingham, Massachusetts, talks about those topics and more. Profit: What can life science companies teach us about the technology for creating healthcare archives? Louie: Life science researcher collaborations share many of the same medical data sharing issues, but because they use anonymous human samples, they have fewer privacy issues and can move faster. Profit: What's the biggest technology challenge? Louie: The huge amount of data associated with medical imaging like MRI, PETs, and CT scans is transforming the IT part of healthcare and life science. Just one small clinical trial involving CT scans on a few dozen people can result in as much as 2 terabytes' worth of data. The same is true for healthcare. The digital storage requirements for hospitals are enormous! Most clinics don't have more than 50 terabytes' worth of storage on hand, but large institutions like Massachusetts General Hospital have 500 terabytes available—and they fill that up every three or four months. Then they send it off to optomagnetic storage and never access it again because it's hard to retrieve. The technology being developed by life science enterprises to access, manipulate, and share these terabytes of data will make scientific research more effective and make healthcare archives possible. Profit: What are the social issues that are preventing a national healthcare archive? Louie: Patient privacy rights are a major concern. When people begin to think about having their medical data collected and put into repositories, it raises thorny issues. Are insurance companies going to discriminate against people because they have a genetic predisposition to a disease? Is someone not going to get a job because a company foresees that his or her healthcare costs will be high? Any national or regional archive associated with healthcare is going to result in these kinds of questions. Profit: Where can we see the technological and social issues being addressed successfully? Louie: A good place to look is the world of biobanking. Biobanking is the pooling of large numbers of tissue samples along with patient medical histories and clinical data so scientists can do analyses, including genetic comparison and gene expression profiling. Sweden's biobank system is among the largest in the world. It requires a robust intranet between member organizations that keeps data secure and accessible. It's not just one group of servers in an isolated facility. They've got to be able to find and update medical data going back several years. They have to link all the repositories together throughout the country. In the United States, the National Cancer Institute is creating a repository of specimens and patient data. Interest groups who share similar health issues often encourage participation in biobanks. So while there are still privacy issues, people are motivated to participate because they want to help scientists find a cure fast. Profit: What steps are being made in the U.S. to protect privacy and set the stage for healthcare archives? Louie: HIPAA, the Health Insurance Portability and Accountability Act of 1996, established the major guidelines for privacy in the U.S. HIPAA implementation is ongoing, and only when people feel sure that HIPAA will protect them will they participate in national archives en masse. Profit: What will a healthcare archive look like once it's built? Louie: Here's one small example: Electronic record keeping will mean a doctor enters the prescription straight into a PDA that connects to a pharmacy over a network that has a quality assurance system in the middle, which says, "Caution: This person shouldn't be taking this medication, because he's allergic." Paperwork will decrease, and the quality of the care will increase.