Structural lesion analysis: applications Dagmar Timmann Department of Neurology University of Duisburg-Essen, Germany
Human cerebellar lesion conditions 1. Focal cerebellar lesions a. Stroke Pro s: - acute testing possible Con s: - mean age 50 years; - chronic: recovery; - acute: disconnection b. Tumors Pro s: - young age Con s: - acute: lesion-symptom mapping not possible; - increased intracranial pressure; - brain development 2. Cerebellar (cortical) degeneration Pro s: - good access; - significant cerebellar signs Con s: - mean age 50 years; - no 100% pure conditions Folie 2 Timmann et al. Neuroscience 2009
Structural MRI sequences Chronic focal lesions (stroke or tumor) 3D T1-weighted: e.g. MPRAGE 1 mm 3 sagittal coronal axial Acute focal lesions (stroke) < 48 hours: diffusion weighted MRI, perfusion MRI Folie 3 > 48 hours: FLAIR
Lesion delineation Gold standard: manual tracing Wilke et al. Neuroimage 2011 ROI Region ( lesion ) of interest www.cabiatl.com/mricro/mricron Folie Rorden 4 and Karnath Nat Rev Neurosci 2004
Normalization Lesions are masked during normalization of cerebellum Brett et al. Neuroimage 2001 SUIT (spatially unbiased atlas template of the cerebellum and brainstem) toolbox in SPM Diedrichsen Neuroimage 2006 SUIT-normalized cerebellum SUIT-normalized ROI ROI overlay on probabilitistic MR atlas Diedrichsen et al. 2009 Folie 5
Lesion symptom maps 1. Superimposition of lesions in patients showing the same disorder 2. Comparing lesion site in two groups of patients - Pooling across predefined anatomical regions - Pooling across behavioural cut-offs - Subtraction analysis 3. Voxel-wise statistical mapping - Binary data - Continuous data Folie 6 www.cabiatl.com/mricro/mricron (including NPM; by Chris Rorden)
Superimposition of lesions in patients showing cerebellar dysarthria n = 31 (surgery) n = 7 (stroke) Lechtenberg and Gilman Ann Neurol 1978 Urban et al. Neurology 2001 Folie 7 No control group, biased by anatomy of lesions
Comparing lesion site in two groups Pooling brain images across predefined anatomical regions Eyeblink conditioning CR CS US Gerwig et al. Brain 2003 Folie 8 1 10 1 12 Posterior inferior Superior cerebellar artery Biased by anatomy of lesions
Comparing lesion site in two groups Pooling brain images based on behavioural cut-offs Working memory (n-back) 3-back 2 6 4 1 6 7 9 8 7.. - Control task: 3-back + sitting - Dual-task: 3-back + tandem gait Folie 9 Ilg et al. in preparation
Comparing lesion site in two groups based on behavioural cut-offs: 90% percentile 3-back + sitting 3-back + tandem gait impaired: 5/17 impaired: 9/17 unimpaired: 12/17 unimpaired: 8/17 Folie 10 http://www.icn.ucl.ac.uk/motorcontrol/imaging/propatlas.htm
Voxel Voxel-wise subtraction analysis Is a given voxel involved in a certain behaviour? impaired Behaviour not impaired with lesion consistent inconsistent no lesion inconsistent consistent (%) consistent voxels (%) inconsistent voxels Folie 11
Voxel-wise subtraction analysis 3-back + sitting Ventral dentate 3-back + tandem gait Dorsal dentate, interposed Cut-off not defined, no inferential statistics done Folie 12 No graduation of abnormality
Inferential statistical mapping - Binary data (no graduation of abnormality) binomial tests: MRICro: chi-square test (Fisher s exact test) NPM in MRICroN: Liebermeister test - Continuous (graduation of abnormality) parametric: t test nonparametric (NPM): Brunner and Munzel test* *Medina et al. Neuropsychologia 2009: at least 10 lesioned and unlesioned - Multiple comparisons: larger groups: permutation smaller groups: FDR Folie 13 T test FDR corrected
T test: p < 0.01, FDR corrected 3-back + sitting: ventral dentate 3-back + tandem gait: dorsal dentate, interposed y=-67 y=-63 y=-59 Folie 14 Ilg et al. in preparation
Problems and future directions Improve lesion delineation Combine data of different MR sequences (e.g. MPRAGE and FLAIR) White matter changes Implement DTI Secondary changes in chronic lesion How to treat secondary atrophy? ROI based normalization of the dentate nuclei acquire 3T or 7T MR images ANOVA: compare two conditions, find possible interactions Folie 15
Cerebellar degeneration Healthy SCA6 Spinocerebellar ataxia type 6 Jung et al. Cerebellum 2011: Region specific atrophy patterns Folie 16
Cerebellar degeneration 1. Conventional volumetry - No normalization required - Predefined anatomical regions 2. Voxel-based morphometry (VBM) - Normalization required (e.g. SUIT) - No predefined anatomical regions - Better spatial resolution Folie 17
Conventional volumetry Reach adaptation Visuomotor Force field Rabe et al. J Neurophysiol 2009 Folie 18
Conventional volumetry Reach adaptation Force field Rotation Visuomotor adaptation and posterior intermediate zone: - L. VI anterior arm area? - Crus I, II parietal reach area? - L. VIII posterior arm area? www.eccet.de Andreas Beck, Volker Aurich University of Düsseldorf Folie 19
VBM: better spatial resolution Reach adaptation Visuomotor Force field L. VI L. V y=-76 y=-62 y=-55 Correlation analysis [t values] Folie Donchin, 20 Diedrichsen et al. J Neurophysiol 2011
Intermediate Reach adaptation Anterior arm area: Subdivision in functional moduls; Force field adaptation L. V Visuomotor adaptation L. VI I-V VI Folie 21 Glickstein et al. 1995, 2009
Problems and future directions Conventional volumetry Do on a lobule wise basis Include volumetry of the nuclei using e.g. SWI T1 mapping instead of VBM White matter changes Implement DTI ANOVA: compare two conditions, find possible interactions Folie 22
Essen Neurology Marcus Gerwig Michael Küper Markus Thürling Neurosurgery Beate Schoch Neuroradiology Elke R. Gizewski Collaborators UCL, London Jörn Diedrichsen Ben Gurion University, Israel Opher Donchin Hertie Institute, Uni Tübingen Winfried Ilg University of Düsseldorf Andreas Beck, Volker Aurich German Research Foundation DFG TI 239/9-1;10-1 Marie Curie Initial Training Grant (EU)