Researchers have announced the achievement of “a century-old aim of neuroscience,” after developing an ultra-accurate map of the human brain that not only more than doubles the number of known cortical regions, but can be reliably reproduced across large numbers of people.
The human brain is a tantalizingly enigmatic organ, and is often said to be the most complicated known structure in the natural world. Architecturally, it consists of an unknown number of regions, each of which differs in its structure, function, and connectivity with other brain areas. Though it is estimated that each hemisphere contains somewhere between 50 and 200 different regions, identifying these has so far proved challenging, as the boundaries between these areas are not clearly defined or easily observable.
At present, our knowledge of the brain’s anatomy is derived from a number of highly limited studies, most of which have attempted to identify cortical regions by examining the brains of small numbers of individuals – often by exploring the brains of dead people.
According to the authors of the latest study – which appears in the journal Nature – most of these previous attempts to delineate the brain’s anatomy have tended to focus on a single property, such as function, connectivity, or architecture. However, by failing to consider all of these simultaneously, the cortical maps that have resulted from these studies have tended to be rather “blurry”, and often don’t line up when applied to the brains of multiple individuals.
To try and address this, the team obtained exceptionally high-quality MRI scans of the brains of 210 people, which were collected using new and refined techniques as part of the Human Connectome Project. These images revealed data regarding cortical thickness, myelin content, and task-related brain activity, acquired by scanning the brains of people as they performed a range of cognitive tasks.
As such, the researchers were able to look at multiple properties in one go, using these to identify 97 cortical regions that were previously not known about. This increases the total number of known cortical regions to 180, representing a massive improvement on the previous figure of 83.
They then programmed a “machine-learning classifier” to recognize the key characteristics of each of these regions, and found that this enabled them to accurately reproduce these maps in a further sample of 210 people, with 96.6 percent accuracy. Amazingly, the computer was even able to locate these cortical regions in individuals with slight brain abnormalities, resulting in some of their brain regions being out of place.
The implications of this research could be hugely significant and far-reaching. For instance, not only could it help advance our knowledge of neuroscience, leading to new insights about cognitive function and mental illness, but it may also help neurosurgeons become more accurate with their scalpels when operating on patients.
On top of this, by examining the structure and function of some of these newly discovered brain regions – particularly those involved with high-level cognition – and comparing these to the neural anatomy of other primates, it’s likely that we’ll get a better idea of how human intelligence evolved.
This is the brain activity of a person listening to a story, with different regions lighting up in different colors according to their patterns of connectivity. Matthew F. Glasser, David C. Van Essen / Nature
