Influence of Multiple Sclerosis Lesions on Magnetic Resonance Image Registration

Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the central nervous system that causes inflammation, demyelination and axonal loss. Magnetic Resonance (MR) imaging presents high resolution and good differentiation between brain tissues and is considered the gold standard for detection and evolution assessment of MS lesions. Registration of MR clinical images (T1/T2/PD/FLAIR) with anatomical brain atlases has proved to be an essential step to automatically detect and segment MS lesions. However, the effect these lesions have on the final results of image registration has not been thoroughly investigated, despite the importance and common use of this pre-processing step in automatic MR image analysis. In this work, image registration techniques using both affine and deformable transformations were analyzed to assess if MS lesions (stratified in mild, moderate or severe) had any effect on the alignment. We introduced misalignments and then registered the images back. This is equivalent to simulating a situation where we want to register a clinical image (misaligned) with a brain template (with no misalignment). Based on quantitative results obtained using volume overlap, Jaccard and Dice, it was verified that MS lesions do not significantly affect the registration process. In the severe lesions case, for instance, the values for the Dice metric were 0.999 and 0.963 for affine and deformable transformation, respectively. Also, it was verified that the sole use of affine transformation is a very reasonable choice to correctly align images even if they do have lesions.