New software aids speed, accuracy of medical image analysis
Physicians who spend inordinate amounts of time and effort analyzing medical images may be able to get much-needed relief from new software that is able to quickly and accurately analyze any series of images from MRIs, CT scans, and ultrasounds, among other diagnostic equipment.
Researchers at the University of Missouri-Columbia developed the software, called Tracking Equilibrium and Nonequilibrium shifts in Data (TREND). They contend that the application, available free for academic use, can significantly reduce the cost and time required to conduct image analysis as well as improve the process of diagnosing patients.
Doctors spend hours, days or even weeks analyzing these kinds of images. However, by using TREND, the researchers contend that analysis takes only a few minutes and is more accurate and consistent than if physicians performed the work themselves.
“It’s as good, or better, than manual analyses,” says Steve Van Doren, a professor in the MU Department of Biochemistry and co-developer of the software. He notes that TREND frees the user from manually choosing which features to monitor, thereby saving time and effort.
In particular, the software resolves the main trends of change in imaging, spectra or other biophysical measurements by using principal or independent component analysis. TREND reads multiple movie formats, PNG images, multiple formats used in nuclear magnetic resonance (NMR) spectroscopy, and data listed in the formats of spreadsheets or text.
“What we’ve actually examined carefully are MRI scans, specifically cardiac MRI movies,” says Van Doren, whose collaborator on the project is Jia Xu, a research scientist in the MU Department of Biochemistry. “We see the strength of this software as being able to resolve the time-dependent changes in the imaging movies quite readily.”
For example, TREND resolves principal components representing breathing and the cardiac cycle in an MRI movie. Using appropriate imaging sources, TREND could be used to quickly detect heart valve defects and arrhythmias, Van Doren says.
The software, which runs on Windows, Mac OS X, and four versions of Linux, is currently free for academic use (registration required). In addition, commercial licensing and a demo version are also available.
“This can be applied to all kinds of medical imaging,” concludes Van Doren. “We’d love to get some clinicians interested in trying it to see how it works and then develop it further to become diagnostically useful.”
More information on TREND can be found here.