Clinicians are able to detect even the earliest signs of cancer or other abnormalities through magnetic resonance imaging (MRI), which scans the inside of the body in intricate detail, however, these scans can be a long and uncomfortable experience for patients as it requires them to lie still in the machine for up to 45 minutes. By using an algorithm developed at MIT’s Research Laboratory of Electronics, scanning times could be lowered to just 15 minutes.
MRI scanners use strong magnetic fields and radio waves to acquire several images of one and the same body part, each designed to create a contrast between different types of tissue. Radiologists are able to detect subtle abnormalities, such as a developing tumor, by comparing multiples images of the same region and examining the variations in contrast of the different tissue types. However, the procedure of taking multiple scans of the same regions is time-consuming, resulting in patients spending prolonged times inside the machine.
Lead author Elfar Adalsteinsson, an associate professor of electrical engineering and computer science and health sciences and technology, and Vivek Goyal, the Esther and Harold E. Edgerton Career Development Associate Professor of Electrical Engineering and Computer Science have developed an algorithm that can dramatically speed up the MRI scanning process. The paper detailing the algorithm will be published in the journal Magnetic Resonance in Medicine.
By using information obtained from the first contrast scan, the algorithm can produce subsequent images without having to start the scanner from scratch each time it produces a different image from the raw data as it already has a basic outline to work from. This considerably shortens the time required to acquire each later scan.
By looking for features that are common in all different scans, such as the basic anatomical structure the software is able to create this outline. The algorithm particularly uses the first scan to predict the likely position of the boundaries between different tissue types in the subsequent contrast scans.
Adalsteinsson explains:
“If the machine is taking a scan of your brain, your head won’t move from one image to the next, so if scan number two already knows where your head is, then it won’t take as long to produce the image as when the data had to be acquired from scratch for the first scan.
Given the data from one contrast, it gives you a certain likelihood that a particular edge, say the periphery of the brain or the edges that confine different compartments inside the brain, will be in the same place.”
According to Goyal the algorithm cannot transfer too much information from the first scan onto subsequent scans because it would risk losing the unique tissue features revealed by the different contrasts.
Goyal explains:
“You don’t want to presuppose too much. So you don’t assume, for example, that the bright-and-dark pattern from one image will be replicated in the next image, because in fact those kinds of dark and light patterns are often reversed, and can reveal completely different tissue properties.”
First author Berkin Bilgic clarifies that the algorithm therefore calculates for each individual pixel what new information is required for constructing the image, and what information, for example the edges of different tissue types it can take from the previous scans. As a result MRI scans are much faster completed and can lower the time that patients have to spend inside the machine from 45 to just 15 minutes. Bilgic admits that the faster scan has a slight impact on the image quality but it is far superior to competing algorithms.
The researchers are currently working on further improving the algorithm so that the raw image data can be processed much faster into a final image that can by analyzed by clinicians once the patients has left the MRI scanner. Standard computer processors take considerably longer for the final step of converting raw data into a final image than conventional MRI scans, however the researchers are optimistic that they are able to reduce the time to the same time required by conventional MRI scans by utilizing recent advances in computing hardware from the gaming industry.
Adalsteinsson comments:
“Graphics processing units, or GPUs, are orders of magnitude faster at certain computational tasks than general processors, like the particular computational task that we need for this algorithm.” He adds, that a student at the laboratory is currently working to implement the algorithm on a dedicated GPU.
Written by: Petra Rattue