Researchers at MIT and Boston Children’s Hospital have created a method to use MRI scans and print physical models of an organ in only a few hours. While 3D printing organ models is not a new technology, the speed of the new method means that surgeons can use the models to plan delicate and time-sensitive surgeries.
The system involves a unique computer algorithm that increases the precision of MRI scans by 10. MIT researchers partnered with Boston Children’s Hospital physicist Medhi Moghari, who created the modeling system, and Andrew Powell, a cardiologist who oversaw clinical work for the project. The team successfully printed a heart model from an MRI, using 10 different patients to test how effective the 3D printed models were.
The team hopes that the models can be used for educational purposes, diagnosing conditions, and helping doctors prepare for surgeries.
“Our collaborators are convinced that this will make a difference,” said leader of the project Polina Golland, a professor of electrical engineering and computer science at MIT. “The phrase I heard is that ‘surgeons see with their hands,’ that the perception is in the touch.”
Prior to this technology, models were printed by manually defining the organ’s boundaries within the MRI scan. The method called for 200 cross sections to ensure precision, and took up to 10 hours to complete.
This study aimed to evaluate the validity of different methods of converting MRI scans to 3D models. Using a computer to define boundaries between different parts of an organ can be problematic, as the distinction between light and dark areas on the MRI scan might not always line up with the actual edges of the anatomical structure.
The researchers found the best results came when using a human expert to pinpoint one-ninth of the boundaries in each of the MRI’s cross sections. After 14 patches, the computer algorithm could infer the remainder of the boundaries with 90% precision for 200 cross sections. Experts who pinpointed all of the boundaries by hand only managed 80% precision.
"I think that if somebody told me that I could segment the whole heart from eight slices out of 200, I would not have believed them," Golland said. "It was a surprise to us.”
Using a combined human expert and computer algorithm to segment sample boundaries takes about one hour, while actually printing the 3D heart takes two more hours.
Next, cardiac surgeons with Boston’s Children’s Hospital will conduct a study to evaluate how useful the models are for medical practice. If the model hearts prove helpful, then 3D printing other organs from MRIs using this method will follow.
MIT and Bryce Vickmark