Abstracts

Rationale: Focal cortical dysplasias (FCD) are malformations of cortical development that often condition medically refractory epilepsy. There is a spectrum of histological and imaging abnormalities with subtle FCDs often overlooked, limiting surgical treatment. Histopathology reveals disorganization of cortical layers, heterotopic neurons and abnormal cells. These features may have limited impact on conventional MR, but can generate considerable signal changes on other modalities. Diffusion-weighted images (DWI) exploit water diffusion to report on microarchitectural characteristics and offer potential for better visualization of FCD. However, the complexity of the cortical tissue warrants advanced diffusion models (Lorio, Epilepsia 2020;61:433–44). We investigated whether advanced diffusion models can identify microarchitectural abnormalities in a rodent model of cortical dysplasia.

Methods: Adult Sprague-Dawley rats were injected with either saline solution (Control) or carmustine (BCNU) i.p. at 14 days of gestation (Benardete, Epilepsia 2002;43:970–82). Rat pups were imaged at 30 postnatal days (n=19 Control, 18 BCNU). Imaging was performed using a 7 T animal scanner. Coronal DWI were acquired in 90 diffusion gradient directions, each with b=670, 1270, and 2010 s/mm² (resolution: 0.175x0.175x1 mm³). Fifteen b=0 s/mm² images were also acquired. Data were analyzed using the tensor model, a multi-tensor model (Coronado-Leija, Med Image Anal 2027;42:26-43), empirical distribution of diffusion and direction distribution (Ferizi, NMR Biomed 2017;30), and spherical mean technique (Kaden, Neuroimage 2016; 139:346-59). To have a common anatomical descriptor of the cortex, we fitted a curvilinear 2D grid to the cortical ribbon in a coronal slice at the level of the primary motor cortex. Diffusion metrics were sampled at 10 equidistant points in each cortical trajectory. For each metric (and at each cortical position), groupwise comparisons were performed. Next, a principal component analysis (PCA) was used to reduce dimensionality; for each point in the cortical mesh we identified the centroid of the control group and evaluated the degree of abnormality of each equivalent point in experimental animals.

Results: Multivariate analysis of all diffusion metrics revealed abnormalities in the majority of the cortex of rats in the BCNU group (Figure 1). Some metrics proved more sensitive to detect group-wise differences. Approaches beyond the diffusion tensor, being able to disentangle diffusion perpendicular or tangential to the pial surface, showed distinct patterns of abnormalities at specific cortical depths (Figure 2).

Conclusions: Advanced diffusion models are able to capture cito- and myelo-architectural information of the normal cortical mantle, and to identify abnormalities of such organization in an animal model of cortical dysplasia. This approach may enhance the detectability of subtle FCD in humans.

Funding: Please list any funding that was received in support of this abstract.: We thank Juan Ortiz-Retana and Nydia Hernández-Ríos for technical assistance. Funding: UNAM-DGAPA (IN204720, IA200621).