Advanced brain diffusion MRI and image texture measures have the potential to predict multi-domain functional outcomes in multiple sclerosis

Background

Multiple sclerosis (MS) causes diverse functional impairments that require early accurate characterization, but pertinent methods are scarce. This study aimed to develop new imaging-driven approaches for predicting MS functions.

New method

19 women with MS [10 relapsing-remitting (RRMS) and 9 secondary progressive (SPMS) subtypes] and 19 matched controls were examined including 3 T imaging. Advanced measures of nerve tract integrity were derived using diffusion MRI and anatomical MRI texture analysis with phase congruency, respectively. Imaging analysis focused on three tract regions critical in MS: the corpus callosum, corticospinal tracts, and optic radiations. Top-ranked tract measures sensitive to MS severity were employed to predict physical, neurocognitive, and affective functions facilitated by Ridge regression.

Results

Top predictors included diffusion apparent fiber density and fractional anisotropy, and phase congruency measures across tracts. The predictions were mostly strong for physical functions including Timed 25-Foot walk, Nine-Hole Peg Test, and neurological disability, strong-to-moderate for neurocognitive functions led by the symbol digit modality test, and relatively weak for affective functions. Further, the normal-appearing white matter (NAWM) models were superior or similar to NAWM+Lesion models based on either imaging type, and the best phase congruency models outperformed the best diffusion-based models.

Comparison with existing methods

Few studies attempted to derive novel measures of nerve integrity using clinical MRI, and virtually no study modelled the utility of these measures for predicting multi-domain MS functions.

Conclusion

Advanced imaging models could predict MS functions for early intervention, especially phase congruency NAWM models for physical functions.