Teaching Watson to 'see'

The goal, say IBM execs is to advance Watson beyond the supercomputer's natural language skills and give it the ability to "see."

IBM researchers estimate that medical images account for at least 90 percent of all medical data today. They also present challenges:

• The volume of medical images can be overwhelming to even the most sophisticated specialists – radiologists in some hospital emergency rooms are presented with as many as 100,000 images a day.
• Tools to help clinicians extract insights from medical images remain very limited, requiring most analysis to be done manually.
• At a time when the most powerful insights come at the intersection of diverse data sets (medical records, lab tests, genomics, etc.), medical images remain largely disconnected from mainstream health information.

IBM plans to leverage the Watson Health Cloud to analyze and cross-reference medical images against a deep trove of lab results, electronic health records, genomic tests, clinical studies and other health-related data sources, already representing 315 billion data points and 90 million unique records.

[See also: CVS, Watson confront chronic disease.]

Merge's clients could compare new medical images with a patient's image history as well as populations of similar patients to detect changes and anomalies. Insights generated by Watson could then help healthcare providers in fields including radiology, cardiology, orthopedics and ophthalmology to pursue more personalized approaches to diagnosis, treatment and monitoring of patients.

Cutting-edge image analytics projects underway in IBM Research's global labs suggest additional areas where progress can be made. They include teaching Watson to filter clinical and diagnostic imaging information to help clinicians identify anomalies and form recommendations, which could help reduce viewing loads for physicians and increase physician effectiveness.

"As Watson evolves, we are tackling more complex and meaningful problems by constantly evaluating bigger and more challenging data sets," Kelly said. "Medical images are some of the most complicated data sets imaginable, and there is perhaps no more important area in which researchers can apply machine learning and cognitive computing. That's the real promise of cognitive computing and its artificial intelligence components – helping to make us healthier and to improve the quality of our lives."

"As a proven leader in delivering healthcare solutions for over 20 years, Merge is a tremendous addition to the Watson Health platform, said John Kelly, senior vice president, IBM Research and Solutions Portfolio, in a news release. "Healthcare will be one of IBM's biggest growth areas over the next 10 years, which is why we are making a major investment to drive industry transformation and to facilitate a higher quality of care,"

"Watson's powerful cognitive and analytic capabilities, coupled with those from Merge and our other major strategic acquisitions, position IBM to partner with healthcare providers, research institutions, biomedical companies, insurers and other organizations committed to changing the very nature of health and healthcare in the 21st century, he added. "Giving Watson 'eyes' on medical images unlocks entirely new possibilities for the industry."

"Today's announcement is an exciting step forward for our employees and clients," added Merge CEO. "Becoming a part of IBM will allow us to expand our global scale and deliver added value and insight to our clients through Watson's advanced analytic and cognitive computing capabilities."

Boon for researchers, clinicians, individuals

IBM's Watson Health unit plans to bring together Merge's product and solution offerings with existing expertise in cognitive computing, population health, and cloud-based healthcare intelligence offerings to:

• Offer researchers insights to aid clinical trial design, monitoring and evaluation;
• Help clinicians to efficiently identify options for the diagnosis, treatment and monitoring a broad array of health conditions such as cancer, stroke and heart disease;
• Enable providers and payers to integrate and optimize patient engagement in alignment with meaningful use and value-based care guidelines; and
• Support researchers and healthcare professionals as they advance the emerging discipline of population health, which aims to optimize an individual's care by identifying trends in large numbers of people with similar health status.