Dentate nuclei (DNs) segmentation is helpful for assessing their potential involvement in neurological diseases. Once DNs have been segmented, it becomes possible to investigate whether DNs are microstructurally affected, through analysis of quantitative MRI parameters, such as those derived from diffusion weighted imaging (DWI). This study developed a fully automated segmentation method using the non-DWI (b0) images from a DWI dataset to obtain DN masks inherently registered with parameter maps. Three different automatic methods were applied to healthy subjects: registration to SUIT (a spatially unbiased atlas template of the cerebellum and brainstem), OPAL (Optimized Patch Match for Label fusion) and CNN (Convolutional Neural Network). DNs manual segmentation was considered the gold standard. Results show that SUIT results have a Dice Similarity Coefficient (DSC) of 0.4907±0.0793 between automatic and gold standard masks. Comparing OPAL (DSC = 0.7624±0.1786) and CNN (DSC = 0.8658±0.0255), showed that a better performance was obtained with CNN. OPAL and CNN were optimised on high spatial resolution data from the Human Connectome Project. The three methods were then used to segment DNs of subjects with Temporal Lobe Epilepsy (TLE) from a 3T MRI research study with DWI data acquired with a coarser resolution. In TLE, SUIT performed similarly, with a DSC = 0.4145±0.1023. OPAL performed worse than using HCP data with a DSC of 0.4522±0.1178. CNN was able to extract the DNs without need for retraining and with a DSC = 0.7368±0.0799. Statistical comparison of quantitative parameters from DWI analysis, as well as volumes, revealed altered and lateralised changes in TLE patients compared to healthy controls. The proposed CNN is a viable option for accurate extraction of DNs from b0 images of DWI data with different resolutions and acquired at different sites.
File in questo prodotto:
Non ci sono file associati a questo prodotto.