overview

Brain Changes in 676 ADNI subjects: Summary of 10 Studiesusing Tensor Based Morphometry& Automated Hippocampal Maps

Paul Thompson and the UCLA ADNI Team*

*Xue Hua, Jon Morra, Alex Leow, Yi-Yu Chou, Suh Lee, April Ho, Christina Avedissian, Sarah Madsen,

Igor Yanovsky, Boris Gutman, Liana Apostolova, Arthur Toga, et al.

*Google “paul thompson ADNI”to get a webpage with links to the full papers (PDFs)

http://www.loni.ucla.edu/~thompson/ADNI/adni.html

http://www.loni.ucla.edu/~thompson/ADNI/adni.html

OverviewMapped brain changes in 676 ADNI subjects

Tensor-based morphometry (gives 3D maps of rates of tissue loss) Automated Hippocampal/Ventricular Mapping 1000s of scans, no manual intervention

Only need ~40 AD and 80 MCI subjects to detect 25% slowing of disease (10x better than best clinical score)

Which MRI measures correlate best with clinical decline, and with CSF biomarkers (A-beta/Tau)?What is the best numeric summary of change from a 3D image? Is 3T better than 1.5T? How is power affected by pooling?

WHAT ARE THE SUBTLEST BRAIN CHANGES WE CAN DETECT?

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Single Subject - Rates of Atrophy mapped with TBM

AD (N=104) versus Normal (N=157)- Mean Atrophy Rates -

Hua, X. and Thompson, P.M. et al., 2009

MCI (N=254) versus Normal (N=157)- Mean Atrophy Rates -


Leow AD, Yanovsky I, Parikshak N, Hua X, Lee S, Toga AW, Jack CR, Bernstein MA, Britson PJ, Ward CP, Borowski B, Trojanowski JQ, Shaw L, Fleisher AS, Harvey D, Kornak J, Schuff N, Alexander GE, Weiner MW, Thompson PM (2009). Alzheimer’s Disease Neuroimaging Initiative: A One-year Follow up Study Correlating Degenerative Rates, Biomarkers and Cognition, NeuroImage, 2009 Apr 15;45(3):645-55.

MCI Converters Lose Tissue Faster

Leow AD, Yanovsky I, Parikshak N, Hua X, Lee S, Toga AW, Jack CR, Bernstein MA, Britson PJ, Ward CP, Borowski B, Trojanowski JQ, Shaw L, Fleisher AS, Harvey D, Kornak J, Schuff N, Alexander GE, Weiner MW, Thompson PM (2009). Alzheimer’s Disease Neuroimaging Initiative: A One-year Follow up Study Correlating Degenerative Rates, Biomarkers and Cognition, NeuroImage, 2009 Apr 15;45(3):645-55.

For drug trials, want to summarize change in a

Region-of-interest (ROI)

Anatomical ROI based on temporal lobes

Statistical ROI derived from an independent training sample of 22 AD patients

Hua, X., Lee, S., Yanovsky, I., Leow, A.D., Chou, Y.Y., Ho, A.J., Gutman, B., Toga, A.W., Jack, C.R. Jr, Bernstein, M.A., Reiman, E.M., Harvey, D., Kornak, J., Schuff, N., Alexander, G.E., Fox, N.C., Weiner, M.W., Thompson, P.M. and the Alzheimer's Disease Neuroimaging Initiative, 2009. Optimizing Power to Track Brain Degeneration in Alzheimer's Disease and Mild Cognitive Impairment with Tensor-Based Morphometry: An ADNI Study of 515 Subjects. To be submitted to NeuroImage.

Statistical ROI reduces sample size by 15-50%; most helpful in MCI

AD MCITBM Designs Stat-ROI Temporal-ROI Stat-ROI Temporal-ROI

sKL-MI S6L8 48 55 88 99sKL-MI S9L5(better for MCI) 52 72 85* 132

Statistically-defined ROI outperforms the anatomically-defined temporal lobe ROI; extremely helpful in MCI, as it focuses on the part of the brain that is changing most*


Sample size estimates for a drug trial (= 48AD, 88 MCI)- does it matter what statistical threshold is used to

define the region with greatest effect sizes for change?


Estimated sample sizes (n80)- needed to detect a 25% reduction in the mean

annual change with a two-sided test and =0.05 at 80% power, for a two-arm study

Sum-of-boxes Clinical Dementia Rating (CDR) gives best power among the clinical scores, but the TBM method is 10 times better

AD MCI

Loss Rate %/yr 48 88CDR-SB 408 796

ADAS-Cog 619 6797MMSE 1078 3275


Is power better at 3T?1.5T

3 T

1.5T average

3 T average

More of the brain showed AD-accelerated tissue loss at 3T than at 1.5 T but with

slightly weaker effect size (24 AD vs. 35 CTLs scanned at both field strengths)

1.5 T 3 THo, AJ., Hua, X., Lee, S., Leow, AD., Yanovsky, I., Gutman, B., Dinov, ID., Lepore, N., Stein, JL., Hojatkashani, C., Toga, AW., Jack, JR., Bernstein, MA., Reiman, EM., Harvey, D., Kornak, J., Schuff, N., Alexander, GE., Weiner, MW., Thompson, PM. Comparing 3 Tesla and 1.5 Tesla MRI for Tracking Alzheimer’s Disease Progression with Tensor-Based Morphometry.

Generated a statistical ROI for each field strength (slightly smaller at 3T)

1.5T ROI (darker gray) slightly larger than 3T ROI (lighter gray)

Ho, AJ., Hua, X., Lee, S., Leow, AD., Yanovsky, I., Gutman, B., Dinov, ID., Lepore, N., Stein, JL., Hojatkashani, C., Toga, AW., Jack, JR., Bernstein, MA., Reiman, EM., Harvey, D., Kornak, J., Schuff, Norbert., Alexander, GE., Weiner, MW., Thompson, PM. Comparing 3 Tesla and 1.5 Tesla MRI for Tracking Alzheimer’s Disease Progression with Tensor-Based Morphometry.

MCI: Power slightly worse at 3T, similar in AD

N80 = Minimal Sample Sizes, per diagnostic group, to detect 25% slowing of the mean atrophic rate (with 80% power, alpha = 0.05).

AD MCI1.5 Tesla 37 107

3 Tesla 48 159Very similar in AD Worse for MCI


0

20

40

60

80

100

120

140

160

180

100% 3T 25% 1.5T 75% 3T 50% 1.5T 50% 3T 75% 1.5T 25% 3T 100% 1.5T

Sample Size Estimate with 80% Power (n80)

ADMCI

Mixing 3T and 1.5T scanners - Power did not degrade at all when

25% of the scanners were 3T

159134 134

103 107

48 52 58

37 37


3T ONLY

1.5T ONLY

MIX IS OK

Morra JH, Tu Z, Apostolova LG, Green A, Toga AW, Thompson PM (2009). Comparison of Adaboost and Support Vector Machines for Detecting Alzheimer’s Disease through Automated Hippocampal Segmentation, IEEE Transactions on Medical Imaging, in press.

HP Loss Rates (980 scans)

Morra, J., Tu, Z., Apostolova, L.G., Green, A., Avedissian, C., Madsen, S., Parikshak, N., Hua, X., Toga, A., Jack, C., Schuff, N., Weiner, M., Thompson, P., 2008. Automated Mapping of Hippocampal Atrophy in 1-Year Repeat MRI Data in 490 Subjects with Alzheimer’s Disease, Mild Cognitive Impairment, and Elderly Controls. Neuroimage.

ApoE4+ atrophied 2-3% faster


AT BASELINE, GROUP DIFFERENCES ARE MAPPED:


Significance Maps

Significance Maps

BASELINE ATROPHY CORRELATES WITH:

Significance Maps

Chou YY, Leporé N, Avedissian C, Madsen S, Parikshak N, Hua X, Trojanowski J, Shaw L, Weiner M, Toga A, Thompson PM (2009). Mapping Correlations between Ventricular Expansion, and CSF Amyloid & Tau Biomarkers in 240 Subjects with Alzheimer’s Disease, Mild Cognitive Impairment & Elderly Controls, NeuroImage. 2009 Feb 21.

Chou YY, Leporé N, Avedissian C, Madsen SK, Parikshak N, Hua X, Trojanowski JQ, Shaw L, Weiner MW, Toga AW, Thompson PM (2009). Mapping Correlations between Ventricular Expansion, and CSF Amyloid & Tau Biomarkers in 240 Subjects with Alzheimer’s Disease, Mild Cognitive Impairment and Elderly Controls, NeuroImage. 2009 Feb 21. [Epub ahead of print]

Summary

All MRI measures correlate well with CSF biomarkers, clinical decline, and predict future conversion to AD.

TBM needs 50 AD and 75 MCI subjects to detect 25% slowing of disease (10x better than best clinical score)

All maps show focal effects - interesting that these statistically guided ROIs will give much better numeric summaries of change (15-50% reductions in sample size)

Mixing 1.5T and 3T scanners is not a problem; but 3T was slightly worse for tracking change in MCI

overview

Documents

normal n

schuff n

parikshak n

tbmad n

mean atrophy rates hua

thompson pm

degenerative rates

d maps of rates