A team of researchers at the Bengaluru-based Indian Institute of Science (IISc) has developed a GPU-based machine learning algorithm that promises to help identify early signs of aging or deterioration of brain function before they manifest behaviourally in Alzheimers patients.
Millions of neurons fire in the brain every second, generating electrical pulses that travel across neuronal networks from one point in the brain to another through connecting cables or “axons”. These connections are essential for computations that the brain performs. Understanding brain connectivity is critical for uncovering brain-behaviour relationships at scale.
The axons are like information highways. Bundles of axons are shaped like tubes, and water molecules move through them, along their length, in a directed manner. A type of scan called diffusion Magnetic Resonance Imaging (dMRI) is used to track these movements. However, the data obtained from the scans only provide the net flow of water molecules at each point in the brain which is not enough to pinpoint the connections.
“Imagine that the water molecules are cars. The information obtained from the scans is just the direction and speed of the vehicles at each point in space and time, but with no information about the roads. The task at hand is similar to inferring the networks of roads by observing these traffic patterns,” explains Devarajan Sridharan, Associate Professor at the Centre for Neuroscience (CNS), IISc, and corresponding author of the study.
Scientists had previously developed an algorithm called LiFE (Linear Fascicle Evaluation) to carry out the work. But, one of its challenges was that it worked on traditional central processing units (CPUs), which made computation time-consuming.
In the new study, Sridharan’s team tweaked their algorithm to cut down the computational effort involved in several ways, thereby improving LiFE’s performance significantly. To speed up the algorithm further, the team redesigned it to work on specialised electronic chips – the kind found in high-end gaming computers – called Graphics Processing Units (GPUs). This helped them analyse data 100-150 times faster than previous approaches.
The improved algorithm named ReAl-LiFE could also predict how a human test subject would behave or carry out a specific task. Using the algorithm, the team could explain variations in behavioural and cognitive test scores across a group of 200 participants.
The researchers noted that such analysis can have medical applications too as data processing on large scales is becoming increasingly necessary for big-data neuroscience applications, especially for understanding healthy brain function and brain pathology.
For example, using their new algorithm, the team hopes to be able to identify early signs of aging or deterioration of brain function before they manifest behaviourally in Alzheimer’s patients.
Besides Sridharan, the study team comprised Varsha Sreenivasan, Sawan Kumar, Partha Talukdar, and Franco Pestilli. They have published a report on their work in the science journal Nature Computational Science.
The study was supported by the Department of Biotechnology-Wellcome Trust India, and the Science and Engineering Research Board, among other funding agencies.
The image shows the superior longitudinal fasciculus (SLF), a white matter tract that connects the prefrontal and parietal cortex, two attention-related brain regions. (Credits: Varsha Sreenivasan and Devarajan Sridharan).
