January 2019: Brain and Development Researchers develop Qoala-T tool
Researchers within the Brain and Development Research Center recently developed the Qoala-T tool, to automatically check the quality of MRI images processed in FreeSurfer. Researchers used to do this by hand, which was subjective, time consuming (may take weeks) and makes it difficult to compare studies between labs. Checking the quality of your data is an highly important step, since image quality issues lead to unreliable results.
As described in their NeuroImage paper (https://doi.org/10.1016/j.neuroimage.2019.01.014), the researcher first rated all processed MRI scans manually (N=784). Next, they used machine learning models to test whether they could then predict the quality of the Freesrufer processed scans. Indeed, manually rated MRI quality could be predicted with [92%] accuracy in this sample and also quite well in two other independent samples.
The research promote others to use the Qoala-T tool, to make the quality control step comparable between studies. As such the source code and model are publicly available online (see https://github.com/Qoala-T/QC). For researchers who easily want to use Qoala-T for their own FreeSurfer-processed scans, there is an online Shiny app that does not require any coding at all. The Qoala-T tool provides a Qoala-T score (ranging from 0 to 100) for every single scan (see https://qoala-t.shinyapps.io/qoala-t_app/). This makes it possible to report average Qoala-T ratings for each dataset.
We really hope that this paper and tool will contribute to replicability of studies by reducing variability in quality control and that many researchers find it useful.