This thesis focuses on the visualisation of cortical multiple sclerosis lesions. First, it examines which MRI sequence can best be used to visualise cortical multiple sclerosis lesions, by comparing five currently existing sequences to each other using histopathology as gold standard. Included MRI-sequences were standard clinical sequences T1, T2, and fluid-attenuated inversion recovery (FLAIR), and more advanced sequences such as double inversion recovery (DIR) and phase-sensitive inversion recovery (PSIR). The results showed that, to detect the highest amount of cortical lesions, a combination of DIR and PSIR sequences should be used. However, these sequences are not readily available in standard clinical and research routines. Therefore, further studies in the thesis focused on generation of artificial DIR images, that were built from readily availableT1- and T2-sequences. These artificially generated DIR images were then assessed for cortical lesions, in a randomized blind setting compared to conventionally acquired DIR images. The results showed that on artificially generated DIR images, almost the same amount of cortical multiple sclerosis lesions can be detected as in conventionally acquired DIR images. However, the sequences that were used to generate the artificial DIR images were all acquired in the same hospital. Therefore, although the results of this study were promising, further implementation of this technique requires multi-centre validation. In the next work in this thesis, multi-centre validation of artificially generated DIR and PSIR images was performed, with experts from seven specialized multiple sclerosis centres around the globe taking part. The results of this multi-centre validation showed that a similar amount of cortical lesions could be detected on artificially generated PSIR images when compared to conventionally acquired images, and an even higher amount of -especially juxtacortical- lesions could be detected on artificially generated DIR images. The final -and crucial- step towards practical implementation of the artificially generated images would be to perform histopathological validation. Thus, artificial DIR images were generated from a cohort of postmortem patients. Two variants of artificial DIR were generated: from T1 and T2 and from T1 and FLAIR. These were then assessed for cortical lesions, compared to conventionally acquired DIR images, in a randomized blinded order. The results showed that when assessing images for cortical lesions without having histopathological feedback at hand, there were no differences in the number of detected lesions between conventionally acquired DIR images and both variants of artificially generated DIR images. With histopathological feedback at hand, a little less lesions were visible on artificial DIR images that were generated from T1 and T2 sequences. The last work in this thesis focuses on the role of cortical lesions in cognitive decline. A cohort of 334 patients with multiple sclerosis underwent MRI and extensive neuropsychological evaluation and was grouped according to isolated cognitive domain decline. Then, associations between cortical lesions and different forms of cognitive decline were assessed, but without significant findings. It could thus be discussed whether cortical lesions can be associated to such specific cognitive deficits, or if they have a more general disruptive character. The main question that remains after this thesis is whether or not to continue on the development of cortical lesion visualisation. As part of the diagnostic criteria, they form an excellent disease marker, but also one of which it is known that 25% can be detected at best. Is it worth large monetary investments to increase this number by 5% whilst we do not know the exact role of these lesions in the disease? Or 10%? This can be discussed from various perspectives and the reader can determine for her- or himself whether she/he thinks if/when Grey Matters should be Revisited once more.
|Qualification||Doctor of Philosophy|
|Award date||22 Feb 2023|
|Place of Publication||s.l.|
|Publication status||Published - 22 Feb 2023|