Worldwide, 1 in 10 people over the age of 65 years and nearly half of people older than 85 years suffer from Alzheimer's disease. The number of people affected with Alzheimer's is expected to double every 20 years.
This progressive disease affects memory, language, thinking, and emotional behavior. At present, there is no cure for Alzheimer's, but certain medications may temporarily improve memory loss and problems with thinking and reasoning. Although the available drugs cannot stop brain cell decline and death, slowing the progression of the disease improves the quality of life for Alzheimer's patients and their caregivers. Such drugs are most useful if patients start taking them early in the disease process. Unfortunately, many patients are diagnosed late, partly due to a lack of specific diagnostic tests for early-stage Alzheimer's. In fact, without a brain biopsy, it is very difficult to know for certain that a person has early-stage Alzheimer's. As brain biopsy is too dangerous to be used for routine diagnosis, researchers are looking for alternative approaches that are noninvasive, yet accurate. Sarang Joshi, associate professor of bioengineering at the University of Utah, is developing mathematical tools and imaging-based markers that one day may allow diagnosis of Alzheimer's at the earliest stages.
What Does an Average Brain Look Like?
Joshi's approach is based on the notion that changes in brain function (e.g., cognitive performance) are accompanied by changes in the shape and size of various areas in the brain. "Going back to Plato and Aristotle, people have been studying how anatomical things vary in a population. The emergence of modern imaging, like MRI, enabled us to ask questions at a much deeper level [about] how the entire shape of anatomy varies," says Joshi.
To understand progression of a disease, one first needs to understand how the average brain is shaped. In the past, researchers have relied on brain atlases based on images from one arbitrarily chosen individual. However, because the shape of the brain varies among individuals in a population, such atlases do not provide a meaningful benchmark for measuring individual anatomical variation. To create meaningful imaging standards, Joshi developed mathematical tools that construct digital three-dimensional atlases of average anatomy by combining brain MRI images from hundreds of individuals. His approach is based on a complex mapping approach, which measures distances on anatomical structures in medical images, allowing researchers to identify patterns of change in brain shape related to disease progression.
Size and Shape Matters in Brain Functioning
In a subsequent study, Joshi sought to identify brain shape deformation patterns that relate to clinical measures of cognitive function, such as word recall, problem solving, and drawing. "We can use shape as a biomarker for disease. That helps with distinguishing one disease from another and distinguishing changes of normal aging from changes associated with disease," says Joshi's collaborator, Richard King, assistant professor of neurology and director of the Alzheimer's Image Analysis Lab at the University of Utah. Researchers have traditionally looked at how the size of certain brain structures, such as ventricles and the hippocampus, changes with progression of dementia and Alzheimer's. But "volume loss in and of itself is not equal to loss of function," indicates King. Joshi's mathematical tools allow researchers to explore more subtle changes in the shape of brain structures that a radiologist could not detect on visual inspection. "We can predict … what an average brain of a particular subject with a particular neuropsych exam looks like," adds Joshi.
Joshi plans to extend this work from structural imaging (e.g., MRI) to functional imaging of the brain (e.g., positron emission tomography, or PET, scanning), which detects and measures physiological activity within a tissue. PET studies have shown that reduced glucose utilization in the brain correlates with severity of cognitive impairment in Alzheimer's and that some patients use different amounts of glucose in their brains' left and right hemispheres. In a recent project, Joshi applied the atlas-building technology to study how this phenomenon, known as metabolic asymmetry, relates to symptoms of cognitive decline in Alzheimer's. Understanding why Alzheimer's causes metabolic asymmetry could contribute to development of new diagnostic and treatment approaches.
Brain Atlases à la Carte
In recent years, Joshi launched AtlasWerks, an open-source software package for construction of custom brain atlases that are used for statistical analysis of brain shape changes. The software is linked to the world's largest Alzheimer's image database. As part of the Alzheimer's Disease Neuroimaging Initiative (ADNI), the database contains brain images, matching demographics (e.g., age, gender, education), and clinical assessments (e.g., dementia rating, audio-visual learning, memory, logical thinking) on about 900 individuals ranging from normal function to mild cognitive impairment to Alzheimer's. "The ADNI database is fully indexed and searchable. We are enabling researchers to define their own criteria and build an atlas for that subpopulation," explains Joshi. For example, one might choose to focus on 60-80-year-old females with mild cognitive impairment. "There are an infinite number of atlases you could make," he adds.
Towards Earlier Detection of Alzheimer's
Analysis of brain shape is currently not used to make clinical decisions. "That's why we need this research so badly," says King. Having quantitative measures would replace the need for subjective, qualitative image assessment, improving diagnostic accuracy. King says the technology may be ready for clinical use in a year or two. But, "the problem is convincing other people to use it. We have to do a lot of work to show people how to use it and show them that it is more effective than what they are currently doing," he adds. His research team is developing MR image analysis tools that complement Joshi's. "Much of the technology that he's developing, I will be trying to use for my approach. It is really great to have colleagues to help with the technology development that I can apply very quickly. These kinds of tools will have immediate impact on my patients' care," says King.
The tools developed by Joshi and King could be useful for early detection of Alzheimer's and other neurological diseases. They might also be used as biomarkers for monitoring disease progression and response to therapy. Joshi plans to interface AtlasWerks with many other neurological disease databases that are being developed. "Now that the infrastructure is in place, we are looking at forming collaborations with people who are studying autism and other diseases," he says.
This work is supported in part by the National Institute of Biomedical Imaging, the National Institute on Aging, and the National Science Foundation.
N, Fletcher PT, Preston JS, Ha L, King R, Marron JS, Wiener M, Joshi S. Multivariate statistical analysis of deformation momenta relating anatomical shape to neuropsychological measures. Med Image Comput Comput Assist Interv. 2010;13(Pt 3):529-37.