Data analysis in brain imaging - Biomedical image analysis group at University of Eastern Finland
This page is permanently bit out of date - check the publications page for the most current information!
As three or four dimensional (3-D, 3-D + time) imaging is the only technique to quantify the brain structure and function in living humans at the system level, analyzing brain imaging data has a unique position within neuroscience. Rapid advances in non-invasive neuroimaging methods have increased the possibilities to study changes occurring in human brain across a variety of time-scales ranging from seconds to entire life span. A large part of these advances can be attributed to the development of dedicated computational algorithms and applied mathematics, which are essential to extract quantitative information from images. My group develops these computational methods to analyze brain imaging data, evaluates the methods, for example, by using advanced simulations, and openly distributes the programs implementing the methods.
In addition to the below projects, I and my group participate to the projects by other groups at our department as collaborators.
Multiscale assessment of epileptogenicity in the human brain – for better diagnosis and treatment of drug-refractory epilepsy. PIs O. Gröhn, A. Sierra, T. Malm, J. Tohka, R. Kälviäinen, funded by Jane and Aatos Erkko foundation, active 2022 - 2025. For individuals with drug-resistant focal epilepsy, the only available treatment option is the surgical removal of the epileptogenic zone. The aim of the project is to develop, firstly, more sensitive and specific MRI approaches for identification of the epileptogenic zone and, secondly, to illuminate what makes tissue epileptogenic at the structural, functional, and molecular level. This is achieved by using novel imaging approaches in pre-surgery MRI. After epilepsy surgery, with the permission of the patients, the dissected tissue is utilized for high-resolution MRI, advanced microscopy and cellular level functional and molecular analysis. As a result, we will create a unique multiscale and multidimensional dataset from individual patients, which will be combined and analyzed using advanced machine learning techniques and by combing data from different sources.
PRIMAL: Robust Framework for PRedictive IMage AnaLysis of Brain Change, PI J. Tohka, funded by the Academy of Finland, active 2022 - 2026. Minute changes in brain structure and function precede visible symptoms in brain diseases. These changes are observable by combining brain imaging and advanced computational data-analysis techniques, specifically within machine learning (ML). This project develops ML methods to facilitate detection of brain changes and to combine imaging information to information on risk-factors and cognitive assessments for personalized decision making concerning brain diseases, such as dementia. The idea of ML algorithms is to use existing data to make computers learn to automatically interpret new data. In this project, a series of brain images and long term follow-up information of the large subject pool form the input to ML algorithms. By utilizing this example data, the algorithms learn to predict disease progression. The use of longitudinal data for prediction is a specific focus area of the project. The developed methods are applied to solve new prediction problems concerning brain health.
Pattern-Cog, PI J. Tohka, funded by EraPerMed, active 2022 - 2025. The EraPerMed funded research consortium Pattern-Cog is composed of six partners from Finland, Germany, Luxemburg, Spain and Sweden, and it is coordinated by the University of Eastern Finland. Pattern-Cog aims to improve dementia prevention strategies by developing and validating a machine learning-based personalized medicine framework for detecting the earliest signs of impending cognitive decline, enabling early and personalized multidomain interventions. J. Tohka is the consortium PI. Please see the Pattern-cog website for current information.
Neuroinnovation PhD programme, co-funded by H2020-MCSA active 2021 - 2026. Neuro-Innovation is a novel and globally unique four-year doctoral programme on research and innovation for brain health throughout life. J. Tohka is a co-PI participating to the supervision of three PhD student in the programme.
Enhancing the innovation potential by advancing the know-how on biomedical image analysis - Kuopio Biomedical Image Analysis Center (KUBIAC) PI. J. Tohka, funded by European Social Fund, 2019 - 2022. Automated image analysis plays an increasingly important role in imaging as the amount of acquired data is continuously growing and traditional, manual image analysis methods start to reach their limits. At the same time, automated image analysis methods have taken giant leaps forward with the development of artificial intelligence (AI) technology, especially in deep neural networks. Methods based on AI will not replace professionals working with the image data, but professionals capable to utilize AI will replace those who are unable to do so. In this project, we will establish Kuopio Biomedical Image Center (KUBIAC) to become a significant producer of image analysis services and related education.
Predictive brain image analysis. PI J. Tohka, funded by Academy of Finland, 2018 - 2022. Minute changes in brain structure and function precede visible symptoms in brain diseases. These changes are observable by combining brain imaging and advanced computational data-analysis techniques. In this project, we develop computational techniques based on machine learning to facilitate the observation of brain changes and diagnose brain diseases earlier. The idea of machine learning algorithms is to use existing data to make computers learn to automatically interpret new data. In this project, brain images and long term follow-up information of the large subject pool form the input to the machine learning algorithms. By utilizing this example data, the algorithms learn to predict disease progression based on brain images. Methods to be developed in this project can be utilized, for example, for early diagnosis of Alzheimer's disease. The project will team up with top level medical experts to try to find most promising utilization targets for the new methods.
Genommed Doctoral Programme . J. Tohka is a co-PI supervising one ESR, co-funded by H2020-MCSA, 2018 - 2022).The GenomMed programme will train international experts in translational genomics aiming to solve health care -related questions in the fields of cardiovascular and metabolic diseases and neurosciences to create solutions and innovations to support development of novel treatments and enhance treatment of patients in the European and global environment.
NeuroDeRisk. J. Tohka is a co-investigator, funded by H2020-IMI2, 2019 - 2022. The adverse effects of pharmaceuticals on the central or peripheral nervous systems are poorly predicted by the current in vitro and in vivo preclinical studies performed during Research and Development (R&D) process. Therefore, increasing the predictivity of the preclinical toolbox is a clear need, and would benefit to human volunteers/patients (safer drugs) and Pharmaceutical Industry (reduced attrition). By combining top level scientists in neurobiology/toxicology with successful software developers, the NeuroDeRisk | Neurotoxicity De-Risking in Preclinical Drug Discovery Consortium will aim at tackling three of the most challenging adverse effects: seizures, psychological/psychiatric changes, and peripheral neuropathies.
Machine learning in brain imaging (PI J. Tohka, active 2015 - 2016, funded by the Universidad Carlos III de Madrid, the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement nº 600371, el Ministerio de Economía y Competitividad (COFUND2013-40258) and Banco Santander.)
This project develops computational methods for machine learning based analysis of large brain image collections. Machine learning refers to the construction and study of methods that can learn models of example data. These models can then be used to make future predictions, for example, predict an early diagnosis for a patient. With these newly developed methods we hope to aid neuroscientists to study on how brain diseases alter the brain structure and function and this way contribute to the better and earlier diagnosis of brain diseases including Alzheimer’s disease, autism, and schizophrenia.
New neuroinformatics methods for automatic analysis of brain images (active 2009 - 2014, funded by Academy of Finland, PI J Tohka)
This project aims at developing automatic methods for the analysis of brain images. This research belongs to the general field of neuroinformatics. Neuroinformatics combines neuroscience and informatics research to develop and apply advanced tools and approaches essential for a major advancement in understanding the structure and function of the brain. The interpretation of world-wide research data would not be possible without new and powerful computational data analysis approaches. The brain research is essential for the treatment of brain disorders and in future also for the prevention of them. The primary goal of neuroinformatics and related brain research is to investigate how a healthy brain functions. Then, by using this knowledge, it will be possible to study disease processes for instance by means of computer simulations.
Laboratory of Neuroimaging, University of Southern California, USA
McConnell Brain Imaging Centre, Monteral Neurological Institute, Canada
Department of Signal Processing, Universidad Carlos III de Madrid, Spain
Reina Sofia Centre for Alzheimer's Research, Madrid, Spain
Oulu Functional Neuroimaging, University of Oulu, Finland
Department of Signal Processing, Tampere University of Technology, Finland
Brain and Mind Laboratory, Department of Biological Engineering and Computational Science, Aalto University, Finland
Strucural Brain Mapping Group, University of Jena, Germany
ITACA institute, Universidad Politécnica de Valencia , Valencia, Spain
Turku PET Centre, Turku, Finland