Dissertation: Data harmonization can improve MRI research
Magnetic resonance imaging (MRI) studies that are conducted in one research site, are constrained by their limited subject and low reproducibility, which significantly impedes the development of MRI research.
Magnetic resonance imaging (MRI) is used in healthcare to diagnose and treat conditions such as stress injuries, prolonged low back pain and neurological conditions.
Pooling data across different research sites, such as Ģֱ and JAMK University of Applied Sciences can improve the statistical power, reliability, and reproducibility of neuroimaging research due to significantly increased sample size.
– However, research site differences represent a barrier when pooling data from different scanners. The existence of the site-noise effect reduces statistical power and can lead to spurious findings. Removing these effects is crucial for successful multi-site data fusion. Additionally, preserving signals of interest is critical in any denoising strategy, says PhD researcher Huashuai Xu.
Xu’s study research introduces a novel dual-projection (DP) approach based on independent component analysis (ICA) that adeptly suppresses site effects while safeguarding the signal of interest.
The ICA-DP method has proven effective in eradicating site-related noises and preserving essential biological variations. This harmonization technique significantly enhances the robustness and accuracy of multi-site MRI data analyses, thereby elevating the trustworthiness of neuroimaging research findings. ICA-DP stands as a valuable tool for future research applications.
– I strongly encourage fellow researchers to integrate the ICA-DP and LICA-DP methodologies into their practices for harmonizing MRI data, paving the way for more consistent and reliable neuroimaging studies, Xu says.
More information
Huashuai Xu defends his doctoral dissertation “Harmonization of multi-site MRI data” 18 December at 12:00.
Opponent is Professor Tarmo Lipping (Tampere University) and custos is Professor Tommi Kärkkäinen (Ģֱ). The language of the dissertation is English.The dissertation can be followed in the lecture hall or online.
.
