Images produced by the technique,
shown in the left column,
are much closer to high-resolution
scans shown in the right column
shown in the left column,
are much closer to high-resolution
scans shown in the right column
MIT and MGH partner to enhance low-quality clinical brain MR images
June 22, 2017
by Lauren Dubinsky, Senior Reporter
A team of MIT researchers and MGH physicians developed a new technique that enhances low-quality MR images of the brain so they can be used for large-scale stroke studies.
More than 795,000 Americans experience a stroke every year, and over 130,000 of the cases are deadly, according to the Centers for Disease Control and Prevention. On top of that, it costs the country an estimated $33 billion each year.
Studying brain MR scans could provide a better understanding of how genetic factors influence stroke survival and how patients respond to different treatments. The challenge is that the resolution is often too low for many analyses.
When stroke patients undergo an MR exam, the images are captured rapidly due to limited scanning time. For scientific studies, researchers keep the subjects in the scanner for much longer in order to obtain higher-resolution images.
“These research studies are very small because you need volunteers, but hospitals have hundreds of thousands of images. Our motivation was to take advantage of this huge set of data,” Adrian Dalca, lead author of a paper on the technique, said in a statement.
The new approach takes information from the entire set of scans and uses it to recreate anatomical features that are missing from individual scans. Once the research-quality images are generated, researchers can run a set of algorithms designed to help with analyzing anatomical features.
During that process, the algorithm keeps track of which pixels came from the original scans and which were filled in afterward, so analyses done later can only be performed on information from the original scans.
Going forward, the MIT team plans to apply this technique to a large set of stroke images obtained by the MGH-led consortium, which includes 4,000 scans from 12 hospitals. They are also hoping to apply it to scans of patients with other brain disorders like Alzheimer’s disease.
“It opens up lots of interesting directions,” said Polina Golland, professor of electrical engineering and computer science at MIT. “Images acquired in routine medical practice can give anatomical insight, because we lift them up to that quality that the algorithms can analyze.”
More than 795,000 Americans experience a stroke every year, and over 130,000 of the cases are deadly, according to the Centers for Disease Control and Prevention. On top of that, it costs the country an estimated $33 billion each year.
Studying brain MR scans could provide a better understanding of how genetic factors influence stroke survival and how patients respond to different treatments. The challenge is that the resolution is often too low for many analyses.
When stroke patients undergo an MR exam, the images are captured rapidly due to limited scanning time. For scientific studies, researchers keep the subjects in the scanner for much longer in order to obtain higher-resolution images.
“These research studies are very small because you need volunteers, but hospitals have hundreds of thousands of images. Our motivation was to take advantage of this huge set of data,” Adrian Dalca, lead author of a paper on the technique, said in a statement.
The new approach takes information from the entire set of scans and uses it to recreate anatomical features that are missing from individual scans. Once the research-quality images are generated, researchers can run a set of algorithms designed to help with analyzing anatomical features.
During that process, the algorithm keeps track of which pixels came from the original scans and which were filled in afterward, so analyses done later can only be performed on information from the original scans.
Going forward, the MIT team plans to apply this technique to a large set of stroke images obtained by the MGH-led consortium, which includes 4,000 scans from 12 hospitals. They are also hoping to apply it to scans of patients with other brain disorders like Alzheimer’s disease.
“It opens up lots of interesting directions,” said Polina Golland, professor of electrical engineering and computer science at MIT. “Images acquired in routine medical practice can give anatomical insight, because we lift them up to that quality that the algorithms can analyze.”