Durante el receso de clases

4. Shared susceptibility model in TLE. As per-vertex FLAIR sig- nal hyperintensity in patients with TLE significantly correlates with per-vertex FLAIR signal covariance to the hippocampus in healthy controls, areas with similar intracortical tissue composition to the hippocampus are the regions that have FLAIR signal pathology in TLE. This may indicate a shared susceptibility to TLE pathol- ogy of regions with similar intra-cortical tissue composition to the hippocampus.

5. Adult TLE patients with a history of febrile convulsions have increased FLAIR signal hyperintensity in the ipsi- lesional parahippocampus. The ipsi-lesional parahippocampus and other transitional cortical regions may be particularly vulnerable to early insults such as febrile convulsions.

6. Surgical outcome is not affected by this extra-lesional pathol- ogy. Neocortical atrophy and FLAIR signal hyperintensity in TLE is not associated with surgical outcome, i.e. the extent of extra- hippocampal abnormalities does not affect chances of seizure free- dom after selective amygdalahippocampectomy.

7. Paediatric HS patients have FLAIR hyperintensities in the ipsi-lesional temporal pole.

8. Cortical thinning in paediatric HS patients is limited to a small area in the ipsi-lesional temporal pole.

9. There is blurring of the grey-white matter boundary in the ipsi-lesional temporal pole in paediatric HS patients. 10. Paediatric HS patients with a history of febrile convulsions

have more pronounced FLAIR changes in the ipsi-lesional temporal neocortex.

6.4

Neurobiological and Clinical Implications

6.4.1

Paediatric FCD

The studies presented in chapters 2 and 3 aimed to address the following questions:

6.4 Neurobiological and Clinical Implications

1. Can characteristics of paediatric FCDs be quantified using surface- based structural MRI analysis?

2. Is machine learning capable of automated detection of FCDs in paediatric epilepsy?

This work demonstrates that surface-based analysis of multi-modal MRI can effectively quantify radiological features of paediatric FCDs. Through the development of novel post-processing methods the sensitivity to characterise and delineate malformations of cortical development can be augmented. Multi-modal MRI and the use of a variety of different features capturing different aspects of cortical morphology and MR signal intensity is crucial given the considerable heterogeneity between individual lesions. This heterogeneity is even present within the same histopathological FCD subtype. Moreover, the work presented in this thesis highlights the ability of machine-learning algorithms to find complex patterns in multivariate data, with a trained artificial neural network able to accurately detect 73% of paediatric FCDs as the first cluster and an additional lesion was detected when the top 5 clusters were analysed.

This work supports the incorporation of automated structural MRI- based techniques into the pre-surgical evaluation of children with drug- resistant focal epilepsy. The crucial question is how to translate these findings into a clinical tool that can be used prospectively in the pre- surgical evaluation of children with epilepsy at Great Ormond Street Hospital and in other epilepsy centres.

6.4.2

TLE in adults

The work presented in chapter 4 aimed to address the following questions: 1. Do T2-weighted signal abnormalities extend beyond the hippocam-

pus in adult epilepsy patients with TLE?

2. Is the topography of neocortical pathology related to hippocampal networks?

3. Do any clinical factors influence the extent of neocortical pathology? This work shows a temporo-limbic topography of FLAIR changes in adults with TLE. The neocortical areas affected are those connected to

6.4 Neurobiological and Clinical Implications

the hippocampus in healthy networks of hippocampal FLAIR intensity co- variance. As FLAIR signal is sensitive to intra-cortical tissue composition, this indicates a susceptibility of neocortex with similar intra-cortical tissue composition to the hippocampus to FLAIR signal abnormalities. Further- more, the finding that FLAIR hyperintensities in these temporo-limbic neocortices are more pronounced in patients with a history of febrile convulsions implies a particular susceptibility of these neocortices to early insults. These findings led to the proposal of the shared susceptibility model in TLE, where areas with similar intra-cortical tissue composition to the hippocampus have a shared vulnerability to pathology. Given that post-mortem histopathological studies have found evidence of grey-matter gliosis in similar neocortical regions to those I report (Blanc et al., 2011; Margerison and Corsellis, 1966) and that FLAIR signal is sensitive to gliosis (Briellmann et al., 2002), it is likely that the increased FLAIR signal intensity found in temporo-limbic regions is caused by an extensive gliotic process. Furthermore, although FLAIR signal is also sensitive to iron and myelin, there is limited evidence for neocortical demyelination (Thom et al., 2000) and decreased iron content (Zhang et al., 2014) in TLE.

6.4.3

Paediatric TLE

The paediatric TLE work aimed to address the following questions: 1. Can quantitative analysis of hippocampal volume and FLAIR in-

tensity lateralise paediatric HS patients?

2. Do morphological or MR signal intensity changes extend beyond the hippocampus?

3. Is there a relationship between any clinical factors and neocortical pathology?

This work demonstrates that quantitative analysis of hippocampal volume and FLAIR signal intensity analysis can help lateralise paedi- atric HS patients. Furthermore, cortical morphological, FLAIR signal intensity and blurring of the grey-white matter boundary are present in the ipsi-lesional temporal lobe. This may stem from dual pathology (i.e. subtle malformations of cortical development in the ipsi-lesional temporal pole) or be due to gliotic or myelin alterations. Further work correlating