Big data & radiology

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - data

Although radiology made tremendous progress in the last two decades in the transition from film to digital, the terabytes of data housed in radiology departments remain stagnant and under-used, according a perspective in the March issue of Journal of the American College of Radiology. However, the capability to transform “dumb” data into knowledge has arrived, with pioneers demonstrating the value of the approach.

“The focus of the next 20 years will be turning dumb data from large and disparate data sources into knowledge and also using the ability to rapidly mobilize and analyze data to improve the efficiency of our work process,” wrote James H. Thrall, MD, chair of radiology at Massachusetts General Hospital (MGH) in Boston.

Thrall cited several examples that elucidate the power of big data. The fusion of decision support with computerized physician order entry drives just-in-time knowledge delivery at the point of care. Possible orders and indications for radiology exams stretch into the thousands; the various permutations extend well beyond the capacity of any physician.

MGH leverages a unique knowledge extraction program (QPID) to mine EHRs, searching for multiple data elements, such as contraindications, simultaneously, and uses these inputs to automatically calculate key figures. QPID aggregates and displays 50 key data elements in the emergency room, highlighting the most important abnormal results and freeing physicians from searching multiple databases, according to Thrall.

Lexicon Mediated Entropy Reduction (LEXIMER) analyzes free-text radiology reports to determine radiologist recommendations. The program could be applied to collections of imaging reports to identify phenotypes and subphenotypes of disease manifestations. LEXIMER also could run concurrently with speech recognition and provide a check to ensure there are no discrepancies in reports.

Radiology Case Tracking (RaceTrack) allows radiologists to enroll patients into the program and then automatically receive updates, such as pathology reports and clinical outcomes, about the patients, thus helping to close the loop between imaging and outcomes.

Finally, one MGH subspecialty division has piloted a dashboard that offers an at-a-glance view of workload and backlogs, enabling redeployment as needed. 

Other big data applications in radiology include automated alerts, computer-aided diagnosis and real-time issues reporting for quality monitoring.

“It is time now to recalibrate our thinking from 'data in' to 'knowledge out' and data mining for work process enhancement. The benefits in safety, quality, cost and stakeholder satisfaction will be enormous,” concluded Thrall.