Artefact reduction in functional MRI

Functional MRI (fMRI) is subject to a number of noise sources such as brain pulsation and subject motion that can substantially impact the quality of results.

To address this we developed an algorithm to automatically and objectively detect artefact arising from noise sources in fMRI, conservatively preserving components of likely neuronal origin, which is essential in clinical applications. The method is fully automated (no training required) and is a self-contained / stand-alone application (no external database access required). We call the algorithm the Spatially Organised Component Klassificator (SOCK).

Whilst SOCK can help reduce the effects of artefacts present in fMRI data, ideally artefacts should also be minimised at the time of data acquisition. For example, we discovered that even a small amount of breath-holding during an fMRI scan can substantially alter results. Our systematic study of this effect alerted investigators to the danger of this potential confound. The problem is best avoided: for example by instructing study participants prior to their scan to breathe normally.