Communications of the ACM
Computers Do Better Than Humans at Measuring Some Radiology Images
Credit: sussexphysio.co.uk
Ohio State University (OSU) researchers have developed software that can analyze images of knee ligaments much faster and just as reliably as humans. Having more information about the meniscus, a blend of fibrous tissue and cartilage in the knee, could lead to its use as a biomarker in predicting those at risk for developing osteoarthritis.
The new method uses several magnetic resonance imaging slices of the meniscus, which can be used to determine changes in its volume over time. Manual interpretation of these images takes between seven and 20 minutes, while the software completes its analysis in two to four minutes.
In developing the program, researchers created algorithms based on the pixel intensity of each image. "The solution is not to get thousands of people to do the work, the solution is to rely on computers," says OSU professor Mark Swanson.
From Ohio State Research News
View Full Article
Abstracts Copyright © 2010 Information Inc., Bethesda, Maryland, USA 
Comments
Razvan Popovici
Congratulation for the achievement, but the article does not evaluate the amount of false positives and false negatives yielded by the software, as compared to the radiologist evaluation. Further, the discrepancy between the actual prevalance of the disease (1/9) and the ratio between affected and healthy cases considered (10/14), corroborated with the small number of experiments (24) may lead to inability to statistically validate the correctness of the prediction tool.
Displaying 1 comment