Researchers: Prof. S. Ramakrishnan and Ranjan Piyush
She was pretty and charismatic. Her chic and poise revealed much about her meticulousness. Or, so said those who knew her from earlier days. Now, she was a picture that was far from pretty; her hair was matted and unkempt, and she needed help with eating, bathing, brushing or just about any other chore. She had seen phases of deterioration – from random memory lapses to frantic calls to dear ones at odd times to total days of reclusiveness. Then came the terrifying days of her believing she was living in a time 20 years ago – recollecting birthdays and funerals alike, wanting to respond to invites that had expired decades ago.
This could, not too far away in the future, be our story. Dementia, a neurodegenerative disorder, is a silent killer. It progressively affects various aspects of one’s daily life and eventually imposes a bedridden state on the afflicted.
Alzheimer’s disease is one of the most common forms of dementia,
affecting an estimated four million people in India.
A person afflicted by Alzheimer’s may not remember having had breakfast, and in the same afternoon, may forget their own identity. Nightmarish as this prospect is, we do not yet have a cure for Alzheimer’s. A cure can mean very different things at different stages of evolution of medical science; halting the disease and postponing its onset are also equally important. Today, the ability to predict the onset of Alzheimer’s is of great interest to researchers and clinicians, with the hope that we may be able to equip impending victims of Alzheimer’s to face the first few years of affliction.
Early diagnosis is, therefore, important. Ranjan Piyush, for his M.S. thesis in the Biomedical Engineering Group of the Department of Applied Mechanics, under the guidance of Prof. S. Ramakrishnan, investigated a possible methodology that may eventually help in early diagnosis of Alzheimer’s.
Although first identified in 1907, we are still far from understanding what causes Alzheimer’s and how to treat it. The structure and function of various regions of the brain and the strength of their inter-linking connections are areas of intense interdisciplinary research. Our understanding of neuroscience has for long been about identifying chunks of the brain or “modules” that are relevant to a certain function.
Today, the tracts, the fibres that connect these modules, have come to be recognized as important. Diffusion Tensor Imaging, or DTI, is an imaging technique that illuminates these connecting tracts and helps us analyze their strength.
When we talk of tickling our gray cells, we are actually referring to the gray matter, which lies on the outer layer of the brain, and is typically associated with cognition, memory and high-level processing. It is composed of cell-bodies of nerve cells, the neurons, that indeed appear gray.
But there is much more to our brains than just the gray regions. A significant volume is occupied by the white matter, composed of axons — spindly fibres along which signals travel — that connect various chunks of the brain. It has a white hue due to the presence of an insulating protein called myelin, which helps prevent leakage and dissipation of the signals. Myelin also enables the signal to travel almost a hundred times faster than it would otherwise.
The white matter of the brain has for long been neglected, but researchers now realize that damage to it, or its sub-optimal functioning, is associated with many neurological conditions like dyslexia, schizophrenia and bipolar disorder, to name a few. Alzheimer’s too is suspected to be closely linked to sub-optimal functioning of the white matter.
Quality of the connections is one of the foremost features of the white matter and DTI is currently the only tool that is in vivo — it can be employed on living, whole organisms, as opposed to one on dead animals or in laboratory petri-dishes. DTI helps us assess the strength of these connections. Cognitive, emotional and social functions of the brain are thought to be deeply associated with the quality of connections.
An understanding of these communication networks, and more importantly,
how certain abnormal or sub-optimal connections are linked to particular conditions,
is a huge challenge in today’s neuroscience research.
Quantitative analysis of DTI images reveals some of the properties of these tracts in the brain. DTI, a modified form of magnetic resonance imaging (MRI), uses the gradient of the magnetic field to probabilistically estimate the diffusion of water molecules in a tissue. We get a vector (technically, a tensor) map of diffusion in various directions, which is dynamic, as the amount — both the magnitude and direction — of diffusion changes in time. This dynamic map is processed to yield the connecting tract-integrity.
Manual analysis of DTI images for diagnosis of Alzheimer’s is time consuming and often does not capture all the information available. So, Ranjan, for his M.S. thesis, developed an automated approach to identify the regions that are most relevant for detecting Alzheimer’s.
Implementing machine learning algorithms that extract features prominently linked to the pathology of Alzheimer’s as well as removing redundant and unwanted ones, is primarily what Ranjan worked on. Tractography, or the analysis of linking tracts, is a vital part of his work. Continuous white matter pathways are constructed from the DTI images, and the average length of tracts is a by-product of this construction.
“To impart a physical touch, we thought, why not measure them [the number of tracts and their length]? And that was the eureka moment. We nailed down on the best available algorithm for tractography and started out to measure the length and number of tracts in various regions of the brain and compare patterns between DTI images of normal and Alzheimer’s subjects’ brains,” says Ranjan.
This comparison revealed a significant difference. The smaller length of tracts associated with Alzheimer’s, he says, may be attributed to loss of neuronal pathways. Qualification of the disorder on the basis of this feature seems a promising technique for diagnosis.
“Our motivation towards working on this particular problem stems from understanding the importance of early diagnosis in this particular disease,” says Ranjan. If we learn to predict the disease in its early stages, we might be able to trace its growth and any accompanying structural, biological and functional changes. Imaging technology is at the core of such research efforts, as engineers try and develop methodologies and algorithms to extract information about neurological conditions from images of the brain.