Optimizing function and imaging measures in multiple sclerosis: the road towards clinical use

Identifying neurological disability and characterizing its magnitude and impact, is of importance for MS management. Although the diagnosis and treatment of MS patients has evolved considerably in the past few decades, identifying neurological deficits through accurate methods that reliably capture the progression of disability is still an area that needs improvement. In part, this is because only imprecise estimates of the clinical and subclinical status of the MS patients’ functioning can be obtained based on existing clinical and para-clinical measures. The aim of this thesis was to optimize existing function and imaging measures in MS patients, and to explore new methods to assess upper extremity function (UEF) and cognition in an advanced way. First, we investigated the assessment of UEF, independently from ambulation, and we explored new psychometric techniques to capture combined clinical judgments of such motor performance in a quantative manner. The ultimate goal here is to establish new innovative techniques for automated quantification of disability through machine learning algorithms (MLAs). Second, we evaluated long-term psychometric properties of two widely applied brief cognitive screening tools in MS, while introducing regression-based norms for the Symbol Digit Modalities Test (SDMT) in a Dutch population. This may eventually facilitate the appropriate assessment of cognitive symptoms in the routine clinical evaluation of MS patients and improve monitoring over time. Finally, we aimed to define how MS pathological changes affect the performance of state-of-the-art algorithms for segmenting the deep grey matter (dGM) on MRI, and how these might influence the relationship between thalamus volume measurements and cognition in MS. We developed a protocol for creating representative manual reference delineations of dGM structures in MS patients, and explored a new semi-automated technique to accelerate such expert outlining. This will provide the opportunity to create larger datasets with expert reference segmentations specifically of MS cases; these can be used in the future, to train MS-specific methods, which we expect to be more accurate when applied to new MS cases than generic methods. Taken together, these three components can lead to early and accurate identification of UEF and cognitive impairment (CI) in patients with MS and subsequently allow for adequate counselling.