Computer Approaches Human Skill for First Time in Mapping Brain
Scientists have for the first time created an algorithm that is almost as good as humans at mapping the neural networks. This could revolutionize our understanding of the brain, and usher in a new era of more effective treatments against diseases like Alzheimer's, Parkinson’s, and so on, where doctors continue to use trial and error methods to treat the patients. In addition to their medical benefits, understanding the human brain in all its complexity is the first step in the direction of augmenting it with technology. The potential in this field is tremendous.
Neural networks are among the most complex and challenging problems before the scientists at the moment. Mapping a neural network is like mapping a distribution of many homes, the roads connecting them, the destinations of various people living in all those homes. Each nerve cell in the human brain shares as much as 7,000 connections with the nerve cells in its vicinity. Mapping them is a gigantic task. Presently, the only way we know how to do it is the manual way.
Scientists use an electron microscope to capture a photo of a few neurons and their circuitry with their neighbors. Then, they manually map the circuitry of all the neurons in the image. The same is done for all the neurons in the brain. These individually developed manual maps are then collated to create a complete picture. This is such a difficult and time-consuming task that it took researchers roughly a decade to just map a single living organism’s brain, and that organism was a worm with only 302 neurons. Compare that with human brain, which has more than 100 billion neurons.
The amount of data that is required to map the human brain is 1000 exabytes, which is the equivalent of all the data that is currently available in the world. Clearly, this isn’t something that humans could do manually.
Enter the research team at Washington State University.
The WSU research time has created an algorithm that maps the neural networks just like the humans do – in multiple stages. The algorithm takes the image of the neural network and processes it in multiple stages, similar to how we do it, with remarkable accuracy that is approaching the accuracy of human level. That being said, the computational model developed by the WSU team is far from perfect. It still commits an enormous number of mistakes.
The WSU’s computational model is still a far cry from the highly accurate model we need to extract any tangible benefits from mapping the human brain. Still, it is a major leap in that direction. Once these algorithms achieve the same level of accuracy as the humans, they will be able to map the entire human brain in only a small fraction of the time than what it would take otherwise.