Automatic segmentation of dentate nuclei for microstructure assessment: example of application to temporal lobe epilepsy patients

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 they are microstructurally affected, through analysis of quantitative MRI parameters, such as the ones derived from diffusion weighted imaging (DWI). This study, therefore, aimed to develop 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 in order to segment the DNs: 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 the segmentation obtained with SUIT has an average Dice Similarity Coefficient (DSC) of 0.4907±0.0793 between the automatic SUIT masks and the gold standard. A comparison with manual masks was also performed for OPAL (DSC = 0.7624 ± 0.1786) and CNN (DSC = 0.8658 ± 0.0255), showing a better performance when using CNN. OPAL and CNN were optimised on heathy subjects’ data with high spatial resolution from the Human Connectome Project. The three methods were further used to segment the DNs of a subset of subjects affected by Temporal Lobe Epilepsy (TLE). This subset was derived from a 3T MRI research study which included DWI data acquired with a coarser resolution. In TLE dataset, SUIT performed similarly to using the HCP dataset, with a DSC = 0.4145 ± 0.1023. Using TLE data, OPAL performed worse than using HCP data: after changing the probability threshold the DSC was 0.4522 ± 0.1178. CNN was able to extract the DNs using the TLE data without need for retraining and with a good DSC = 0.7368 ± 0.0799. Statistical comparison of quantitative parameters derived from DWI analysis, as well as volumes of each DN, revealed altered and lateralised changes in TLE patients compared to healthy controls. The proposed CNN is therefore a viable option for accurate extraction of DNs from b0 images of DWI data with different resolutions and acquired at different sites.

[1]  Chuan Zhou,et al.  Deep Learning in Medical Image Analysis. , 2017, Advances in experimental medicine and biology.

[2]  Jörn Diedrichsen,et al.  A spatially unbiased atlas template of the human cerebellum , 2006, NeuroImage.

[3]  Daniel C. Alexander,et al.  SANDI: A compartment-based model for non-invasive apparent soma and neurite imaging by diffusion MRI , 2019, NeuroImage.

[4]  Wilfried Philips,et al.  MRI Segmentation of the Human Brain: Challenges, Methods, and Applications , 2015, Comput. Math. Methods Medicine.

[5]  Michael Seidenberg,et al.  Cerebellar atrophy in temporal lobe epilepsy , 2005, Epilepsy & Behavior.

[6]  Matthew J. Betts,et al.  The whole-brain pattern of magnetic susceptibility perturbations in Parkinson’s disease , 2017, Brain : a journal of neurology.

[7]  Christoph Meinel,et al.  Deep Learning for Medical Image Analysis , 2018, Journal of Pathology Informatics.

[8]  A. Alexander,et al.  Diffusion tensor imaging of the brain , 2007, Neurotherapeutics.

[9]  I. Soltesz,et al.  Cerebellar Directed Optogenetic Intervention Inhibits Spontaneous Hippocampal Seizures in a Mouse Model of Temporal Lobe Epilepsy , 2014, eNeuro.

[10]  Julien Cohen-Adad,et al.  Spinal cord grey matter segmentation challenge , 2017, NeuroImage.

[11]  J. Helpern,et al.  MRI quantification of non‐Gaussian water diffusion by kurtosis analysis , 2010, NMR in biomedicine.

[12]  Steen Moeller,et al.  The Human Connectome Project: A data acquisition perspective , 2012, NeuroImage.

[13]  V. Chan‐Palay,et al.  The Cerebellar Dentate Nucleus , 1977 .

[14]  Jelle Veraart,et al.  Evaluation of the accuracy and precision of the diffusion parameter EStImation with Gibbs and NoisE removal pipeline , 2018, NeuroImage.

[15]  J. Ashburner,et al.  Connectivity derived thalamic segmentation in deep brain stimulation for tremor , 2018, NeuroImage: Clinical.

[16]  Christian S. Perone,et al.  Spinal cord gray matter segmentation using deep dilated convolutions , 2017, Scientific Reports.

[17]  D. Timmann,et al.  Cerebellar pathology in Friedreich's ataxia: Atrophied dentate nuclei with normal iron content , 2014, NeuroImage: Clinical.

[18]  Shunxing Bao,et al.  Using deep learning for a diffusion-based segmentation of the dentate nucleus and its benefits over atlas-based methods , 2019, Journal of medical imaging.

[19]  Jörn Diedrichsen,et al.  Imaging the deep cerebellar nuclei: A probabilistic atlas and normalization procedure , 2011, NeuroImage.

[20]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[21]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[22]  Yaniv Assaf,et al.  Composite hindered and restricted model of diffusion (CHARMED) MR imaging of the human brain , 2005, NeuroImage.

[23]  Essa Yacoub,et al.  The WU-Minn Human Connectome Project: An overview , 2013, NeuroImage.

[24]  Mohammed Bennamoun,et al.  A Guide to Convolutional Neural Networks for Computer Vision , 2018, A Guide to Convolutional Neural Networks for Computer Vision.

[25]  Francesco Visin,et al.  A guide to convolution arithmetic for deep learning , 2016, ArXiv.

[26]  Durga Ram Sure,et al.  Duus’ Topical Diagnosis in Neurology: Anatomy, Physiology, Signs, Symptoms, 5th edition , 2013 .

[27]  C. Habas.  Functional Imaging of the Deep Cerebellar Nuclei: A Review , 2010, The Cerebellum.

[28]  Sébastien Ourselin,et al.  Generalised Wasserstein Dice Score for Imbalanced Multi-class Segmentation using Holistic Convolutional Networks , 2017, BrainLes@MICCAI.

[29]  Jerry L. Prince,et al.  Fully automatic segmentation of the dentate nucleus using diffusion weighted images , 2012, 2012 9th IEEE International Symposium on Biomedical Imaging (ISBI).

[30]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[31]  P. Lantos,et al.  Cerebellar Dentate Nucleus in Alzheimer’s Disease with Myoclonus , 1999, Dementia and Geriatric Cognitive Disorders.

[32]  Lin Chen,et al.  Multi-atlas tool for automated segmentation of brain gray matter nuclei and quantification of their magnetic susceptibility , 2019, NeuroImage.

[33]  T. Babb,et al.  Fastigiobulbar and dentatothalamic influences on hippocampal cobalt epilepsy in the cat. , 1974, Electroencephalography and clinical neurophysiology.

[34]  V. Chan‐Palay,et al.  Cerebellar Dentate Nucleus , 1977, Springer Berlin Heidelberg.

[35]  David Bonekamp,et al.  Diffusion tensor imaging in children and adolescents: Reproducibility, hemispheric, and age-related differences , 2007, NeuroImage.

[36]  I. Mavroudis,et al.  Dendritic, Axonal, and Spinal Pathology of the Purkinje Cells and the Neurons of the Dentate Nucleus After Long-Term Phenytoin Administration , 2013, Journal of child neurology.

[37]  Adam Finkelstein,et al.  PatchMatch: a randomized correspondence algorithm for structural image editing , 2009, SIGGRAPH 2009.

[38]  Neil Davey,et al.  Cerebellar output controls generalized spike‐and‐wave discharge occurrence , 2015, Annals of neurology.

[39]  Marco Catani,et al.  Visualization of the deep cerebellar nuclei using quantitative T 1 and ρ magnetic resonance imaging at 3 Tesla , 2007, NeuroImage.

[40]  Elnaz Jahani Heravi,et al.  Guide to Convolutional Neural Networks , 2017 .

[41]  D. Louis Collins,et al.  An Optimized PatchMatch for multi-scale and multi-feature label fusion , 2016, NeuroImage.

[42]  U. Klose,et al.  Pattern of Cerebellar Atrophy in Friedreich’s Ataxia—Using the SUIT Template , 2019, The Cerebellum.

[43]  Daniel C. Alexander,et al.  NODDI: Practical in vivo neurite orientation dispersion and density imaging of the human brain , 2012, NeuroImage.