NUS scientists take advantage of machine learning to discover new perspectives in the human brain



IMAGE: Assistant professor Thomas Yeo of the National University of Singapore led an interdisciplinary research team to discover new insights into the cellular architecture of the human brain.
watch Plus

Credit: National University of Singapore

An interdisciplinary research team led by scientists from the National University of Singapore (NUS) has successfully employed machine learning to discover new insights into the cellular architecture of the human brain.

The team demonstrated an approach that automatically calculates brain parameters using data collected from functional magnetic resonance imaging (fMRI), which allows neuroscientists to infer the cellular properties of different regions of the brain without exploring the brain using surgical means. This approach could potentially be used to evaluate the treatment of neurological disorders and develop new therapies.

"The underlying pathways of many diseases occur at the cellular level, and many pharmaceutical products operate at the microscale level." To know what actually happens at the innermost levels of the human brain, it is crucial for us to develop methods that can delve into the depths. brain in a non-invasive way, "said badistant professor Thomas Yeo, team leader, who belongs to the Singapore Neurotechnology Institute (SINAPSE) at NUS, and the Clinical Imaging Research Center A * STAR-NUS (CIRC).

The new study, conducted in collaboration with researchers from the Netherlands and Spain, was published online for the first time in a scientific journal. Scientific advances On January 9, 2019.

Unravel the complexity of the human brain.

The brain is the most intricate organ of the human body, and is made up of 100 billion nerve cells that, in turn, are connected to another 1,000. Any damage or illness that affects even the smallest part of the brain could cause serious deterioration.

Currently, most studies on the human brain are limited to non-invasive approaches, such as magnetic resonance imaging (MRI). This limits the examination of the human brain at the cellular level, which can offer new insights into the development and potential treatment of various neurological diseases.

Different research teams from all over the world have taken advantage of biophysical models to close this gap between the non-invasive image and the cellular understanding of the human brain. Biophysical models of the brain could be used to simulate brain activity, allowing neuroscientists to obtain a view of the brain. However, many of these models are based on overly simplistic badumptions, as, for example, all regions of the brain have the same cellular properties, which scientists have known to be false for more than 100 years.

Building virtual brain models.

Prof. Yeo and his team worked with researchers from the Universitat Pompeu Fabra, Universitat Barcelona and the University Medical Center of Utrecht to badyze the image data of 452 participants of the Human Connectome Project. Starting from the previous modeling work, they allowed each region of the brain to have different cellular properties and automatic learning algorithms exploited to automatically estimate the parameters of the model.

"Our approach achieves a better fit with the real data, and we found that the microscale model parameters estimated by the machine learning algorithm reflect how the brain processes information," said Dr. Peng Wang, who is the first author. from the article. , and had done the study when he was a postdoctoral researcher on the team of Asst Prof Yeo.

The research team discovered that regions of the brain involved in sensory perception, such as vision, hearing and touch, exhibit cellular properties opposite to the regions of the brain involved in thought and internal memories. The spatial pattern of the cellular architecture of the human brain closely mirrors how the brain processes hierarchically information about the environment. This form of hierarchical processing is a key feature of both the human brain and recent advances in artificial intelligence.

"Our study suggests that the brain's processing hierarchy relies on microscale differentiation between its regions, which may provide more clues about advances in artificial intelligence," said Asst Prof Yeo, who also works in the Department of Electrical Engineering. and Computing. at the NUS Engineering School.

Next steps

In the future, the NUS team plans to apply its approach to examine the brain data of individual participants, to better understand how individual variation in the cellular architecture of the brain can be related to differences in cognitive abilities. The team hopes that these latest results can be a step towards the development of individualized treatment plans with specific medications or brain stimulation strategies.


Resignation: AAAS and EurekAlert! they are not responsible for the accuracy of the press releases published in EurekAlert. through contributing institutions or for the use of any information through the EurekAlert system.

Source link