Part III Statistical Characterization of Brain Structures via
Part III Statistical Characterization of Brain Structures via M-reps Guido Gerig Departments of Computer Science and Psychiatry UNC Chapel Hill Guido Gerig, UNC, Feb. 2003
Representative Clinical Study: Neuropathology of Schizophrenia • When does it develop ? • Fixed or Progressive ? • Neurodevelopmental or Neurodegenerative ? • Neurobiological Correlations ? • Clinical Correlations ? • Treatment Effects ? Noninvasive neuroimaging studies using MRI/f. MRI to study morphology and function Guido Gerig, UNC, Feb. 2003
Study: Structural analysis of caudate nucleus in Schizophrenia Processing Steps • Automatic whole brain tissue classification (EM segm. ) • User-operated masking of caudate on label image (intra-, interrater reliability > 0. 95) • Surface parametrization of caudate shapes SPHARM & PDM • Alignment/Normalization: Surface Correspondence • Medial mesh generation (m-rep model) Guido Gerig, UNC, Feb. 2003
Modeling of Caudate Shape PDM M-rep Surface Parametrization Guido Gerig, UNC, Feb. 2003
Basal Ganglia Netter’s Atlas of Human Anatomy Ventricles: lateral ventricle 3 rd ventricle temporal horn caudate nucleus hippocampus Guido Gerig, UNC, Feb. 2003
Caudate Shape Analysis • Clinical Groups: CNTL – Healthy controls (N=30) – Typical drug treatment (30) – Atypical drug treatment (30) • Clinical questions: Atyp – Shape difference between groups? – Drug/patient interaction? – Location & type of changes Typ Mean Shapes per Group Guido Gerig, UNC, Feb. 2003
Caudate volume analysis • Significantly larger volumes of SZ versus controls • Trend but not significant difference between Typ/Atyp • Where and what is difference? Guido Gerig, UNC, Feb. 2003 *preliminary analysis, not controlled for age
Mean Shapes CNTL vs. SZ left right Overlay of aligned (transl/rot) original shapes: green: CNTL / purple mesh: SZ Guido Gerig, UNC, Feb. 2003
Mean Shapes CNTL vs. SZ left right Overlay of size normalized* shapes: green: CNTL / purple mesh: SZ Guido Gerig, UNC, Feb. 2003 *shape should not reflect size change
Alignment, Correspondence? – Choice of alignment coordinate system? – Establishing correspondence is a key issue for building statistical shape models. – Various methods for definition of correspondence exist (landmarks, high dimensional warping, PDM w. MDL refinement, …). – Resulting eigenmodes of deformation depend on these definitions. – Scaling of objects prior to shape analysis? Guido Gerig, UNC, Feb. 2003
Object Alignment before Shape Analysis 1 stelli TR, no scal 1 stelli TR, vol scal side top side Guido Gerig, UNC, Feb. 2003 Procrustes TRS
Shape Representation Method: Medial Representation M-rep Implied Surface Skeletal Mesh (sampled) Local Width (Radius) Implied shape represents original shape with 99% volume overlap and =0. 05 MAD at boundary (M. Styner, Ph. D thesis) Guido Gerig, UNC, Feb. 2003
Shape Difference Analysis of M-rep Mesh Position Local Width Mesh distance at corresponding nodes: Grp A Object deformation, Bending Grp B A and B aligned, superimposed Guido Gerig, UNC, Feb. 2003 Radius difference at corresponding nodes: Local width change
Statistical Analysis • Shapes represented by m-rep: Significant feature reduction, multiscale • Still: Number of features sample size. • Variability hides shape changes. • Shapes not represented by scalar values: Standard MANOVA analysis inappropriate. • Often: PCA on features, selection of small # of Eigenmodes, Fisher linear discriminant, leave one out test for classification. • But: Fisher LD not robust, #of features? , feature selection? , does PCA reflect group differences? Guido Gerig, UNC, Feb. 2003
Statistical Analysis mesh distance radius Guido Gerig, UNC, Feb. 2003
Guido Gerig, UNC, Feb. 2003
Permutation Test • Monte Carlo Sampling (m=k=30) • Mean differences from 1000 permutations • Test original difference 22. 8 versus distribution: p=0. 025 #experiments Guido Gerig, UNC, Feb. 2003
Results Caudate Shape Analysis Integrated local effects • • *Non-parametric permutation test • Guido Gerig, UNC, Feb. 2003 Typical group shows larger shape difference to controls than atypical group Significant shape difference between typical and atypical treatment group Shape distances not shown in combined SZ versus controls analysis Treatment effect or clinical selection bias? Experimental study design, result need to be verified in cross-validation study
Results Caudate Shape Analysis Comparison of Surfaces Right: CNTL – Typ Right: CNTL – Atyp • • Right: CNTL – Atyp -Typ • Significant shape changes mostly in the head of the caudate Shape effect on left side larger than on right side Local significance tests in progress Guido Gerig, UNC, Feb. 2003 Right: Atyp – Typ CNTL Atyp Typ
Shape Difference: Where and What? Local Mesh Deformation Atypical versus Typical drug treatment groups (N = 30) Local Deformation (Euclidean dist. between corresponding nodes) Local significance tests (nonparametric permutation tests) mesh with node differences p<0. 01 p-values per mesh node mesh with nodes p<0. 05 Guido Gerig, UNC, Feb. 2003
Shape Difference: Where and What? Local Mesh Deformation Atypical versus Typical drug treatment groups (N = 30) Local Deformation (Euclidean dist. between corresponding nodes) Group A mesh with nodes p<0. 05 Group B Guido Gerig, UNC, Feb. 2003
Shape Difference: Where and What? Local Width Difference Atypical versus Typical drug treatment groups (N_atyp=N_typ = 30) Local Width Diff. (Radius diff. between corresponding node positions) Local significance tests (nonparametric permutation tests) mesh with node differences p<0. 05 p-values per mesh node mesh with nodes p<0. 05 Guido Gerig, UNC, Feb. 2003
Shape Difference: Where and What? Local Width Difference Atypical versus Typical drug treatment groups (N_atyp=N_typ = 30) Local Width Diff. (Radius diff. between corresponding node positions) Group A Group B mesh with nodes p<0. 05 Guido Gerig, UNC, Feb. 2003
Shape Difference: Where and What? Local Width Difference Atypical versus Typical drug treatment groups (N_atyp=N_typ = 30) Local Width Diff. (Radius diff. between corresponding node positions) Morphing between Atypical (thinner) and Typical (thicker) Guido Gerig, UNC, Feb. 2003
Discussion Caudate Study Netter’s Atlas of Human Anatomy Width and mesh deformation mostly in caudate body/head. Secondary mesh deformation posteriorly Typical treatment group differs from Controls, but not Atypical. Clinical implications? Study caudate shape change relative to neighboring shapes. Guido Gerig, UNC, Feb. 2003
Study: Hippocampal Shape in Schizophrenia • IRIS: Tool for interactive image segmentation. • Manual contouring in all orthogonal sections. • 2 D graphical overlay and 3 D reconstruction. • Hippocampus segmentation protocol (following Duvernoy). • Hippocampus: reliability >0. 95 intra-, >0. 85 interrater) Guido Gerig, UNC, Feb. 2003
Hippocampal Volume Analysis Left smaller than right l SZ smaller than CNTRL, both left and right l Variability SZ larger than CNTL l Guido Gerig, UNC, Feb. 2003
3 D Shape Variability: Left Hippocampus of 90 Subjects Guido Gerig, UNC, Feb. 2003
Hippocampal Shape Analysis left right Left and right hippocampus: Comparison of mean shapes CNTL-SZ (signed distance magnitude relative to SZ template) in out Left Right Movie: Flat tail: SZ, curved tail: CNTL Guido Gerig, UNC, Feb. 2003
Hippocampus M-rep: Global & Local Statistical Analysis Hippocampus: Integrated difference to template shape (structures size normalized) Width (p<0. 75) SZ CNTL individual mrep local group discrimination statistics Deformation (p<0. 0001) SZ CNTL Guido Gerig, UNC, Feb. 2003 p<0. 01 G. Gerig & M. Styner
Local Statistical Tests Medial representation study confirms: Hippocampal tail is region with significant deformation. Guido Gerig, UNC, Feb. 2003
Statistical Analysis of M-rep representations • *Work in progress Keith Muller, Emily Kistner, M. Styner, J. Lieberman, G. Gerig, UNC Chapel Hill • Systematic embedding of interaction of age, duration of illness and drug type into local statistical analysis • Correction for multiple tests *Repeated measures ANOVA, cast as a General Linear Multivariate Model, as in Muller, La. Vange, Ramey, and Ramey (1992, JASA). Exploratory analysis included considering both the "UNIREP" Geisser-Greenhouse test and the "MULTIREP" Wilks test. Difference in hippocampus shape between SZ and CNTRL as measured by M-rep deformation M-rep 3 x 8 mesh Tail Guido Gerig, UNC, Feb. 2003 Head
Model: Row x Col x Drug (y/n) x Age: p = 0. 0097 Patient-CNTL Deformation Difference at Age 30 AGE Deformation at mesh nodes (mm) Patient-CNTL Deformation Difference at Age 40 Patient-CNTL Deformation Difference at Age 20 Tail Head Difference in hippocampus shape between patients and controls: Located mostly in the tail of the hippocampus, becomes more pronounced over time. Guido Gerig, UNC, Feb. 2003
Comparison to CNTLs Deformation at mesh nodes (mm) Change in hippocampus shape over ten years for controls Tail Head Guido Gerig, UNC, Feb. 2003
Conclusions • Shape represents changes not reflected by volume analysis • Several clinical studies: Shape discriminates better than volume • M-rep superior to boundary models – separate analysis of local width/bending – results explained in natural language terms – potential to analyze figure-subfigure relationships and figures in anatomic context • Improved statistical framework for discrimination in development Guido Gerig, UNC, Feb. 2003
Acknowledgements • • • Martin Styner Sean Ho Sampath Vetsa Keith Muller Jeffrey A. Lieberman Stephen M. Pizer and M-rep team Talk at: http: //midag. cs. unc. edu Guido Gerig, UNC, Feb. 2003
- Slides: 36