Developing tools and services to combine datasets obtained by different medical centers
The brain can store more information than a supercomputer. It can process a flood of information about the world around us by seeing, hearing, smelling, tasting, and touching. The brain has enabled humans to achieve breathtaking milestones — walking on the moon, mapping the human genome, and composing masterpieces of art, literature, and music. But brain disorders, in particular neurodegenerative diseases and mental illnesses, are also among the most prevalent and debilitating diseases of our time.
Neuroscientists are specialized in the study of the brain. Over the years, the neuroscience field has made enormous progress. Scientists continue to strive for a deeper understanding of the workings of the brain, and the causes of specific brain diseases.
By analyzing neuroimaging data, neuroscientists can find biomarkers (measurableindicatorof some biological state or condition) that predict who is at risk of developing an illness. The volume of the hippocampus, for example, has huge potential as a marker for Alzheimer’s disease. The larger the sample size, the better the predictive value of the biomarker, which can be achieved by combining data sets obtained by different medical centers. This requires an eScience infrastructure for standardized image analysis, and the exchange of imaging data, meta-data, and analysis results.
The aim of the Biomarker Boosting project is to build a reusable platform for sharing patient (subject) imaging data among hospitals to run a common analysis pipeline. This considerably increases the number of datasets available for analysis, and therefore greatly improves the statistical value of the results.
The multidisciplinary team is developing tools and services for cooperative research in finding better biomarkers and demonstrating its potential by applying it to four different cohorts collected to study dementia and mild cognitive impairment (MCI). For this purpose a common catalogue (meta-data in which definitions of database objects are stored) is designed to allow for data exchange using web services, together with a template data sharing agreement to allow access to researcher-initiated consortia. The second focus is to implement a standardized image analysis pipeline to calculate biomarker features.
The product of the analysis pipeline is a set of low-dimensional properties derived from the high-dimensional images. These properties are called biomarkers, with ‘volume of the hippocampi’ as an example. The improved statistical power is called boosting, hence the title ‘biomarker boosting’.
This project will demonstrate progress in clinical neuroscience through the use of eScience by developing technologies to
a) merge highly heterogeneous, distributed data collections
b) allow for secured and regulated access to the data and its derivatives
c) establish a framework for biomarker extraction.
Image: US Air Force (CC License)