1. Baseline and longitudinal patterns of brain atrophy in MCI patients, and their use in prediction of Short- term conversion to AD: Results from ADNI Xuejiao Chen 2012.12.7

2.  Introduction  Materials and methods  Results  Discussion and conclusion Outline

3.  Author: Chandan Misra, Yong Fan, Christos Davatzikos  Published in NeuroImage Introduction

4.  Goal:  Utilize high-dimensional pattern classification of baseline scans to determine predictors of short-term conversion from MCI to AD.  Measure the spatial distribution of brain atrophy, as well as its longitudinal change, in MCI converters (MCI-C) and in MCI non-converters (MCI-NC) in the ADNI cohort, and to evaluate differences between these two groups. Introduction

5.  Hypothesis  Advanced pattern analysis and classification techniques applied to baseline and longitudinal images of the regional distribution of brain tissues would allow us to predict future conversion from MCI to AD. Introduction

6.  Introduction  Materials and methods  Results  Discussion and conclusion Outline

7.  ADNI  800 adults, ages 55 to 90  200 CN older individuals followed for 3 years  400 MCI followed for 3 years  200 early AD followed for 2 years Materials and methods

9.  Image analysis  Pre-processing:  Alignment to the AC-PC plane  Skull-stripping  Tissue segmentation into GM, WM, and CSF  High-dimensional image warping to a standardized coordinate system, a brain atlas was aligned with the MNI coordinate space  Formation of RAVENS maps using tissue-preserving image warping. Materials and methods

10.  RAVENS maps quantify the regional distribution of GM, WM, and CSF, since one REVENS map is formed for each tissue type.  RAVENS values in the template’s (stereotaxic) space are directly proportional to the volume of the respective structures in the original brain scan. Materials and methods

11.  In order to minimize longitudinal “jitter noise”  Aligned all scans of each individual to his/her baseline scan Materials and methods

12.  Statistical analysis and pattern classification  Group comparison performed via voxel-based statistical analysis of respective REVENS maps normalized by intracranial volume, smoothed using 8mm FWHM smoothing kernel.  Group comparisons involved voxel-by-voxel t-test applied by SPM to the RAVENS maps and to voxel-wise regression maps obtained from longitudinal RAVENS measurements. Materials and methods

13.  Statistical analysis and pattern classification  Group comparisons of the rates of change in the RAVENS maps were calculated by first applying voxel- wise linear regression to all available RAVENS maps of each participant, and subsequently applying SPM’s t-test. Materials and methods

14.  Statistical analysis and pattern classification  Classifier:  A classifier was constructed from the CN and AD groups, and then applied to the MCI subjects.  A more specific classifier that optimally differentiates between MCI-C and MCI-NC was constructed and tested using leave-one-out cross-validation. Materials and methods

15.  Introduction  Materials and methods  Results  Discussion and conclusion Outline

16.  Voxel-based analysis of baseline RAVENS maps  GM: anterior hippocampus, amygdala, temporal lobe GM, insular contex, posterior cingulate, orbitofrontal cortex  WM: periventricular frontal region, right hippocampus  CSF: temporal horns of the lateral ventricles Results

19.  Voxel-based analysis of regression maps of rate of RAVENS change  No significant differences in the rages of change of GM  A relatively faster increase of periventricular abnormal WM  Higher rates of CSF increase in MCI-C Results

21.  Pattern classification between MCI-C and MCI-NC (Classifier 2)  Maximum classification rate: 81.5%, AUC=0.77 Results

23.  Introduction  Materials and methods  Results  Discussion and conclusion Outline

24.  Significant differences at baseline were measured between MCI-C and MCI-NC.  MCI-C had reduced GM volumes in a number of brain regions and parahippocampal WM.  Larger size of the temporal horns of the ventricles.  MCI-C had already reached levels of widespread and significant brain atrophy at baseline. Discussions and conclusion

25.  Used individual patient classification using pattern recognition methods (Classifier 1).  Scores of AD-like patterns of brain atrophy were determined for each individual.  Almost all MCI-C effectively had AD-like brains at baseline. Discussions and conclusion

26.  High-dimensional patterns classification (Classifier 2) indicated that a relatively reasonable classification accuracy can be obtained.  A number of MCI-NC have AD-like structural profiles, which indicates that these individuals might become converters soon. Discussions and conclusion

27.  Higher level of periventricular abnormal tissue, which appears gray in T1 images and is known to indicate small-vessel disease.  Measuring vascular pathology is bound to be very important in predicting which individuals will convert to AD. Discussions and conclusion

28.  Only significant group differences in the longitudinal rate of change were found in measures of periventricular abnormal WM and the temporal horn’s CSF volumes.  Practically obtaining robust measurements of rate of change from a limited number of imaging measurements is problematic  variations in scanner performance and tissue contrast  noise in the image analysis measurements Discussions and conclusion

29.  Though the ADNI data had been corrected for scanner distortions, the baseline measurements are far better markers of patterns of atrophy distinguishing converters from non-converters.  Baseline atrophy is the cumulative effect of possibly many years of progression of AD pathology. Discussions and conclusion

30.  The use of high-dimensional pattern classification methods also helped amplify the predictive value of baseline measurements.  Rate of change is nonetheless a significant biomarker in clinical studies evaluating treatment effects. Discussions and conclusion

31.  Investigated the use of a high-dimensional pattern classification approach as a means to obtain a sensitive and specific biomarker of AD-like spatial patterns of brain atrophy, and of conversion from MCI to AD.  The results of this short-follow-up study are encouraging and indicate that subtle and spatially complex patterns of brain atrophy can be detected and quantified with relatively high accuracy. Discussions and conclusion