Researchers devise elegant algorithms for developing brain mapping with MRI – ScienceDaily

Scientists at Japan’s brain science project have used machine intelligence to improve the accuracy and reliability of a powerful brain mapping technique, a new study reports.

Their development, published on 18 December in Scientific Reports, giving researchers more confidence in using this method to solve human brain wiring and gain a better understanding of the changes in this wiring that are associated with brain or mental disorders such as Parkinson’s disease or Alzheimer’s.

“Working out how the different brain regions are connected – what we call the brain’s connectome – is crucial to fully understanding the brain and the complex processes it undergoes,” said Professor Kenji Doya, who heads the Neural Computation Unit at the Graduate University of the Okinawa Institute of Science and Technology (OIST).

To identify connectomes, researchers monitor nerve cell fibers that extend throughout the brain. In animal experiments, scientists can introduce a fluorescent trace to several points in the brain and image where the cloud fibers coming from those points expand. But this process requires the analysis of hundreds of brain chips from many animals. And because it is so aggressive, it cannot be used in humans, explained Dr. Doya.

However, advances in magnetic resonance imaging (MRI) have made it possible to estimate noninvasive connections. This technique, called MRI-based fiber tracking, uses powerful magnetic fields to track signals from water molecules as they move – or spread – across nappies. A computer algorithm then uses these water signals to estimate the path of the cloud fibers throughout the entire brain.

But for now, the algorithms are not yielding definite results. Just as images may look different depending on the position of the camera chosen by a photographer, the options – or parameters – that scientists choose for these algorithms can generate different connections. .

“There are real concerns about the reliability of this approach,” said Dr. Carlos Gutierrez, the first author and postdoctoral researcher in the OIST Neural Computing Unit. “False objects can be under the control of the links, meaning they show cloud connections that don’t really exist.”

In addition, the algorithms struggle to detect nerve fibers that stretch between remote regions of the brain. But those long-distance connections are some of the most important for understanding how the brain works, Dr. Gutierrez said.

In 2013, scientists launched a Japanese government-led project called Brain / MINDS (Brain Mapping with Integrated Neurotechnologies for Disease Studies) to map the brains of marmosets – small non-human primates with a similar structure human brain.

The Brain / MINDS project aims to create a complete connection of the marmoset brain using both a noninvasive MRI imaging technique and the invasive fluorescent detection method.

“The data set from this project was a great opportunity for us to compare the results from the same brain created by the two approaches and decide what parameters need to be set to make the most accurate connection. based on MRI generation, “said Dr. Gutierrez.

In the current study, the researchers tried to widely use the parameters of two different algorithms so that they could reliably detect long threads. They also wanted to ensure that the algorithms identified as much timber as possible and identified those that were not present.

Instead of trying all the different parameter combinations manually, the researchers turned to machine information.

To determine the optimal parameters, the researchers used an evolution algorithm. The fiber tracking algorithm estimated the connectome from the transmission MRI data using parameters that were altered – or suppressed – in each subsequent generation. These parameters competed against each other and the best parameters – the ones that generated the closest connections to the neural network discovered by the fluorescent detector – were passed on to the next generation.

The researchers tested the algorithms using fluorescent detections and MRI data from ten different marmoset brains.

But it was not easy to choose the best parameters, even for tools, the researchers found. “Some parameters may reduce the false positive rate but make it more difficult to find long-range connections. There is a conflict between the various issues we want to resolve. So deciding which parameters to choose is each. always involving trade, “said Dr. Gutierrez.

Through the many generations of this “survival-as-appropriate” process, the algorithms running for each brain changed the optimal parameters of each other, allowing the algorithms to settle on a set of more likely parameters. At the end of the process, the researchers took the best parameters and averaged them to create a single shared set.

“Combining parameters was an important step. Individual brains change, so there will always be a specific combination of parameters that works best for one particular brain. But Our goal is to come up with the best set of common parameters that would work well for all marmoset brains, “explained Dr. Gutierrez.

The team found that the algorithm with the standard set of fully developed parameters also generated a more accurate link in a new marmoset brain that was not part of the original training set, compared to the standard parameters used. previous use.

The striking contrast between the images taken with algorithms using the basic and fully developed parameters sends a strong warning about MRI-based connectome research, the researchers said.

“It raises the question of any study using undeveloped or proven algorithms,” he warned Dr. Gutierrez.

In the future, the scientists hope to make the process of using machine information to identify the best parameters faster, and use the improved algorithm to link brains with disorders. determine brain or mind more accurately.

“Ultimately, MRI-based fiber tracking could be used to map the entire human brain and identify the differences between healthy and diseased brains,” Dr. Gutierrez said. “This could take us one step closer to learning how to handle these disorders.”