The results of this study – a collaboration between researchers from Montreal’s CHU Sainte-Justine Children’s Hospital and publishedTS Engineering School – are published today in Frontiers in Neuroscience.
“This is the first time that artificial intelligence has been used to better define various parts of a newborn’s brain on an MRI: namely gray matter, white matter, and cerebrospinal fluid,” Dr. Gregory A. Lyodgensky, a neonatologist, said CHU Sainte-Justine and professor at the Universito de Montréal.
“Until today, the tools available were complex, often esoteric and difficult to use,” he said.
In collaboration with Professor Jose Dolez, an expert in medical image analysis and machine learning at AndTS, researchers were able to adapt and then validate the instruments to the specifications of the neonatal setting.
This new technology allows the brain of infants to be tested quickly, accurately and reliably. Scientists see this as a major asset to support research that not only addresses brain development in neonatal care, but also the effectiveness of neuropatrodiic strategies.
In evaluating a range of devices available in artificial intelligence, CHU Sainte-Justine researchers found that these devices had limitations, particularly in relation to pediatric research. Today’s neuroimaging analysis programs are designed primarily to work on “adult” MRIs. The brain immaturity of newborns complicates such analysis, with the inverse of contrasts between gray matter and white matter.
Inspired by Dolz’s most recent work, researchers proposed an artificial neural network, which learns how to efficiently combine information from multiple MRI sequences. This method made it possible to automatically better define various parts of the brain in the newborn and establish a new criterion for this problem.
“We have decided to share the results of our study not only on open source, but also computer code, so that brain researchers everywhere can take advantage of it, which all patients take advantage of,” said Dolz.
CHU Sainte-Justine is one of the most important players in Canada’s neonatal brain platform and is also one of the largest neonatal units in Canada specializing in neurodevelopment. As part of the forum, research teams are implementing projects aimed at improving the long-term health of those newborns who are most vulnerable to brain injury.
“In studies to assess the positive and negative effects of various treatments on the maturation of children’s brains, we have the ability to quantify brain structures with certainty and reliability.” “By offering the fruit of all our discoveries to the scientific community, we are helping them, at least for newborns, creating an extraordinary benefit.”
He said: “Now we want to democratize this tool so that it becomes the criterion for studying brain structure in newborns around the world. For this, we are working on its generality – that is, its use on MRI data in various hospitals. ”
On March 26, 2020, Frontiers in Neuroscience was published. Gregory A. The first author, under the direction of Leodengski, is Yang Ding, Ph.D. The primary author is Gregory A.
Lodygensky, MD, Clinical Associate Professor in the Department of Pediatrics at the Université de Montréal and a Clinician-Researcher at CHU Sainte-Justine, and Assistant Professor in the Department of Software Engineering and Information Technology at the Dolcole de technologie supérieure, Jose Dolz, PhD. ÉTS). The study was supported by the Brain Canada Foundation.
Deep learning neural networks are particularly powerful in dealing with structured data, such as pictures and volumes.
Both the modified LiviaNET and HyperDense-Net performed well in a prior competition with 6-month-old infant magnetic resonance images, but newborn neurological tissue type identification has been challenged, reflecting its uniquely inverted tissue contrasts . The aim of the present study is to evaluate two architects who divide the types of neonatal brain tissue at the same age.