Biomedical image analysis
Current research lines:
Machine learning for brain imaging: Our major research focus lies in identifying biomarkers of brain disorders from neuroimaging data, which is an exciting and rapidly growing research area at the intersection of machine learning, biomedical engineering and neuroscience.
Conventional approaches towards imaging biomarkers reduce the data dimensionality by averaging the image information to one or few variables of a-priori interest - for example, the volume of Hippocampus for Alzheimer’s diagnosis. However, such methods discard much information present in brain images. Instead, allowing machine learning algorithms to decide what is important and decipher the predictive pattern (sometimes called “statistical biomarker”) is projected to be beneficial. This leads to challenging and underconstrained machine learning problems where the data dimensionality is larger than the number of samples and advanced computational techniques are required to solve these problems. We develop these techniques and apply to them to large brain image databases to help neuroscientists to find imaging markers to different brain disorders.
Inter-subject correlation based analysis of fMRI: This project develops and validates statistical and computational methods to analyze fMRI data using inter-subject correlation (ISC). ISC methodology is based on voxel-wise correlation between the time series of the subjects, which makes it completely non-parametric and thus suitable for naturalistic stimulus paradigms such as movie watching. The implementations of the developed methods will be made available under open access ISC toolbox https://www.nitrc.org/projects/isc-toolbox/
More information: www.jussitohka.net