How about spending minutes, not hours, tuning CT and MR scan data for 3D models of a patient's anatomy?

When MIT Media Lab's Steven Keating, Ph.D., then a 26-year-old grad student at its Mediated Matter group, found he had a brain tumor, now safely removed, he grew curious to see his own brain before his surgery to better understand what he had and the therapy options he faced.

He collected all his scans and tried to prepare them for printing, but grew frustrated with the tools at his disposal, which were cumbersome and inaccurate.
So he reached out to his lab colleagues, who were researching new ways to print 3D models of biological samples.

"It never occurred to us to use this approach for human anatomy until Steve came to us and said, 'Guys, here's my data, what can we do?'" says Ahmed Hosny, who was a Research Fellow at the Wyss Institute at the time and is now a machine learning engineer at the Dana-Farber Cancer Institute.

A loose collaboration followed, including scientists at Wyss, as well as researchers and physicians at centers in the U.S. and Germany, that has now developed a novel technique to easily and quickly convert medical images into models with heretofore unattained detail, they reported in the journal 3D Printing and Additive Manufacturing.

"I nearly jumped out of my chair when I saw what this technology is able to do," recalled co-author Dr. Beth Ripley, assistant professor of radiology at the University of Washington and clinical radiologist at the Seattle VA. "It creates exquisitely detailed 3D-printed medical models with a fraction of the manual labor currently required, making 3D printing more accessible to the medical field as a tool for research and diagnosis."

The problem is that the volumes of data from imaging like MR and CT are loaded with so much detail that the points of interest can get lost. This requires that you highlight the things you want to see to distinguish it from surrounding tissue – a very time-intensive process called "segmentation" in which a radiologist must actually trace the objects of interest on every single slice, by hand.

The alternative is automatic "thresholding", in which a computer converts grayscale pixels into either solid black or solid white pixels, depending on a specified “threshold” between black and white.

Unfortunately, since medical data has many ill-defined borders between objects, both computers and hand methods tend to over- or under-exaggerate features and lose vital details.

CT Homepage