pith. sign in

arxiv: 1907.06064 · v1 · pith:YRWRBSZWnew · submitted 2019-07-13 · 📡 eess.IV · cs.CV· cs.LG· stat.ML

Image Evolution Trajectory Prediction and Classification from Baseline using Learning-based Patch Atlas Selection for Early Diagnosis

Can Gafuroglu , Islem Rekik This is my paper

Pith reviewed 2026-05-24 21:54 UTC · model grok-4.3

classification 📡 eess.IV cs.CVcs.LGstat.ML
keywords MRIpatch-based analysisevolution trajectory predictionearly mild cognitive impairmentatlas selectionsupervised learningmanifold learningclassification
0
0 comments X

The pith

Selecting similar baseline MRI patches lets researchers average their observed future changes to predict a new scan's trajectory and classify early mild cognitive impairment.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops supervised and unsupervised ways to choose which training atlas patches at a single baseline timepoint best match a given test patch. Once chosen, the method retrieves the real follow-up trajectories of those atlas patches and averages them to forecast how the test patch will evolve. The predicted trajectories are then passed to an ensemble of landmark-specific linear classifiers. This matters because clinical datasets for early mild cognitive impairment often lack multiple timepoints, so the approach aims to turn one scan into a usable forecast of disease progression. The reported result is that classification accuracy rises by as much as 10 percentage points over methods that use only the baseline image.

Core claim

By training bidirectional mappings from pairwise patch similarities to prediction errors (supervised) or by embedding baseline atlas and test patches in a multi-kernel manifold (unsupervised), the best atlas patches can be identified; averaging their recorded evolution trajectories then yields a predicted trajectory for the test patch that improves downstream classification.

What carries the argument

Learning-based patch atlas selection that maps baseline similarity or manifold position to the patches whose observed trajectories are averaged to forecast a test patch's evolution.

If this is right

  • An ensemble of linear classifiers, each trained at one brain landmark, can label the predicted trajectories as normal control or early mild cognitive impairment.
  • Both the supervised similarity-to-error mapping and the unsupervised multi-kernel manifold can be used to perform the atlas selection step.
  • The method directly targets the scarcity of longitudinal MRI acquisitions in early mild cognitive impairment studies.
  • Classification performance improves by up to 10 percentage points relative to single-timepoint baselines.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If baseline similarity reliably predicts trajectory similarity, the same selection logic could be tested on other progressive brain conditions where follow-up data are limited.
  • Successful averaging of trajectories implies that local intensity patterns at one timepoint carry information about regional atrophy rates.
  • The unsupervised manifold version might extend to multi-modal images without retraining the supervised error model.

Load-bearing premise

That patches similar at baseline will have similar future evolution trajectories that can be averaged to predict a new patch accurately.

What would settle it

A held-out test set in which the averaged trajectories from baseline-selected atlas patches produce classification accuracy no higher than using the baseline patch alone.

Figures

Figures reproduced from arXiv: 1907.06064 by Can Gafuroglu, Islem Rekik.

Figure 1
Figure 1. Figure 1: Illustration of the proposed patch-specific evolution trajectory prediction and classification framework from a baseline MRI using supervised (A) and unsupervised (B) strategies. (A) Supervised atlas patch selection strategy. We first learn a mapping function f t1 i at each landmark xi , which aggregates two bidirectional regressors f +t1 i and f −t1 i (see Section II for more details), to map the intensit… view at source ↗
Figure 2
Figure 2. Figure 2: Modeling the similarity between a baseline atlas patch p t1 i,s0 of subject s centered at landmark i and a second baseline atlas patch p t1 i,s for the proposed supervised atlas selection (SAS) strategy. We give an example of two training atlas patches with varying similarities to an input baseline training patch. We can clearly see how the quotient vector αs,s0 (resp., αs,s00 ) locally captures the degree… view at source ↗
Figure 3
Figure 3. Figure 3: Positive and negative disparity matrices construction. Given two training baseline patches {p t1 i,s, p t1 i,s0}, both centered at landmark i acquired at baseline timepoint t1 for different subjects s and s 0 , we compute their element-wise bi-directional differences: d + s,s0 and d − s,s0 , each representing a row vector in the positive and negative disparity matrices, respectively. In this figure, we giv… view at source ↗
Figure 4
Figure 4. Figure 4: Image evolution trajectory prediction and classification results. (A) and (C): Average predicted patch accuracy evaluated using mean absolute error (MAE) for each ROI in the left and right hemisphere respectively, using the proposed strategies for predicting the follow-up image evolution trajectory from baseline. (B) and (D): The average classification accuracy of our proposed methods for each ROI in the l… view at source ↗
Figure 5
Figure 5. Figure 5: Patch prediction results at two representative landmarks in the left ventricle using supervised atlas selection (SAS) method. The red arrow points to bright voxels where the residual between the ground truth patch and the predicted patch was large. across ROIs, which we opted for in this paper for comparative training at a fixed spatial scale. Also, we note that for SAS the direction of the error influence… view at source ↗
read the original abstract

Patients initially diagnosed with early mild cognitive impairment (eMCI) are known to be a clinically heterogeneous group with very subtle patterns of brain atrophy. To examine the boarders between normal controls (NC) and eMCI, Magnetic Resonance Imaging (MRI) was extensively used as a non-invasive imaging modality to pin-down subtle changes in brain images of MCI patients. However, eMCI research remains limited by the number of available MRI acquisition timepoints. Ideally, one would learn how to diagnose MCI patients in an early stage from MRI data acquired at a single timepoint, while leveraging 'non-existing' follow-up observations. To this aim, we propose novel supervised and unsupervised frameworks that learn how to jointly predict and label the evolution trajectory of intensity patches, each seeded at a specific brain landmark, from a baseline intensity patch. Specifically, both strategies aim to identify the best training atlas patches at baseline timepoint to predict and classify the evolution trajectory of a given testing baseline patch. The supervised technique learns how to select the best atlas patches by training bidirectional mappings from the space of pairwise patch similarities to their corresponding prediction errors -when one patch was used to predict the other. On the other hand, the unsupervised technique learns a manifold of baseline atlas and testing patches using multiple kernels to well capture patch distributions at multiple scales. Once the best baseline atlas patches are selected, we retrieve their evolution trajectories and average them to predict the evolution trajectory of the testing baseline patch. Next, we input the predicted trajectories to an ensemble of linear classifiers, each trained at a specific landmark. Our classification accuracy increased by up to 10% points in comparison to single timepoint-based classification methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript proposes supervised and unsupervised frameworks to jointly predict and classify the evolution trajectory of intensity patches seeded at brain landmarks from baseline MRI data for early eMCI diagnosis. Atlas patches are selected via bidirectional mappings from pairwise similarities to prediction errors (supervised) or multi-kernel manifold embedding (unsupervised); their observed trajectories are averaged to predict the test patch trajectory, which is then fed to an ensemble of per-landmark linear classifiers. The central empirical claim is an accuracy increase of up to 10 percentage points over single-timepoint classification methods.

Significance. If the trajectory prediction step is shown to be accurate and the accuracy gain is attributable to it rather than ancillary design choices, the work could address the scarcity of longitudinal scans by enabling early diagnosis from baseline data alone. The patch-atlas and ensemble-classifier design is a plausible way to capture subtle, heterogeneous atrophy patterns. Credit is due for explicitly targeting the single-timepoint limitation and for combining supervised error-driven selection with an unsupervised multi-scale manifold approach.

major comments (3)
  1. [Abstract (method description)] The 10-percentage-point accuracy gain is presented as arising from the trajectory-prediction step, yet the manuscript provides no quantitative test (e.g., prediction error on held-out longitudinal patches or correlation between baseline similarity and trajectory similarity) that patches close at baseline remain close in their longitudinal change. This assumption is load-bearing for attributing the gain to the proposed averaging mechanism rather than to ensemble construction or landmark choice.
  2. [Abstract (results claim)] No experimental details are supplied on cohort size, number of timepoints, cross-validation folds, exact baseline comparators, statistical testing of the 10-point gain, or rules for data exclusion. Without these, it is impossible to evaluate whether the reported improvement survives multiple-comparison correction or post-hoc selection bias.
  3. [Abstract (supervised technique)] In the supervised bidirectional-mapping procedure, training on the mapping from similarity to prediction error risks circularity: the selection criterion is derived from the same error quantity later used to evaluate the averaged trajectory, so any reported gain may partly reflect the training objective rather than genuine out-of-sample trajectory fidelity.
minor comments (2)
  1. [Abstract] Typo: 'boarders' should read 'borders'.
  2. [Abstract (unsupervised technique)] The description of how the multi-kernel manifold embedding produces patch selection is too terse to reproduce; a short algorithmic outline or pseudocode would help.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed report. We address each major comment below. Where the comments identify gaps in validation or clarity, we agree to revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract (method description)] The 10-percentage-point accuracy gain is presented as arising from the trajectory-prediction step, yet the manuscript provides no quantitative test (e.g., prediction error on held-out longitudinal patches or correlation between baseline similarity and trajectory similarity) that patches close at baseline remain close in their longitudinal change. This assumption is load-bearing for attributing the gain to the proposed averaging mechanism rather than to ensemble construction or landmark choice.

    Authors: We agree that a direct quantitative check of the core assumption would strengthen attribution of the observed gain to the atlas-selection and averaging step. The current experiments report only end-to-end classification accuracy. In revision we will add (i) mean-squared prediction error of the averaged trajectories on held-out longitudinal patches and (ii) Spearman correlation between baseline patch similarity and trajectory similarity, computed on the training folds. These metrics will be reported alongside the classification results. revision: yes

  2. Referee: [Abstract (results claim)] No experimental details are supplied on cohort size, number of timepoints, cross-validation folds, exact baseline comparators, statistical testing of the 10-point gain, or rules for data exclusion. Without these, it is impossible to evaluate whether the reported improvement survives multiple-comparison correction or post-hoc selection bias.

    Authors: The full manuscript contains these details (ADNI cohort of 200+ subjects, 2–3 timepoints per subject, 10-fold CV, comparison against single-timepoint SVM and random-forest baselines, paired t-tests with Bonferroni correction). However, the abstract is too terse. We will expand the abstract to include cohort size, CV scheme, and statistical test, and we will add a supplementary table listing exact exclusion criteria and p-values for the accuracy differences. revision: yes

  3. Referee: [Abstract (supervised technique)] In the supervised bidirectional-mapping procedure, training on the mapping from similarity to prediction error risks circularity: the selection criterion is derived from the same error quantity later used to evaluate the averaged trajectory, so any reported gain may partly reflect the training objective rather than genuine out-of-sample trajectory fidelity.

    Authors: The mapping is learned exclusively on training-set pairs; for each test patch the learned regressor predicts an error for every training atlas patch without ever seeing the test patch’s own follow-up data. The final accuracy is therefore measured on truly held-out subjects. We nevertheless recognize that an explicit statement of this train/test separation would remove any ambiguity. We will add a clarifying paragraph and a flowchart in the methods section. revision: partial

Circularity Check

0 steps flagged

No circularity: selection mapping trained on atlas pairs then applied to test data

full rationale

The paper describes training bidirectional mappings on atlas-atlas baseline similarities to their observed prediction errors, then selecting atlas patches for a test patch via the learned mapping and averaging the selected atlas trajectories. This pipeline uses held-out atlas data for training the selector and applies it to new test patches; the final averaged trajectory is not equivalent to the test input by construction. No equations, self-citations, or uniqueness claims are present that reduce the claimed prediction to a renaming or fit of the inputs themselves. The derivation remains self-contained as a standard supervised selection plus averaging procedure.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract provides insufficient detail to enumerate specific free parameters or invented entities; central claim rests on domain assumption that baseline similarity implies similar future trajectories.

axioms (1)
  • domain assumption Evolution trajectories of baseline-similar atlas patches are representative for averaging to predict test patch trajectory
    Invoked in the step where selected atlas trajectories are averaged after patch selection

pith-pipeline@v0.9.0 · 5848 in / 1164 out tokens · 35444 ms · 2026-05-24T21:54:21.145737+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

45 extracted references · 45 canonical work pages · 1 internal anchor

  1. [1]

    Alzheimer’s disease and vascular dementia in developing countries: prevalence, management, and risk factors,

    R. N. Kalaria, G. E. Maestre, R. Arizaga, R. P. Friedland, D. Galasko, K. Hall, J. A. Luchsinger, A. Ogunniyi, E. K. Perry, F. Potocnik, M. Prince, R. Stewart, A. Wimo, Z.-X. Zhang, and P. Antuono, “Alzheimer’s disease and vascular dementia in developing countries: prevalence, management, and risk factors,” The Lancet Neurology, vol. 7, no. 9, pp. 812 – 8...

    work page 2008

  2. [2]

    Mild cognitive impairment represents early-stage alzheimer disease,

    M. JC, S. M, M. J, and et al, “Mild cognitive impairment represents early-stage alzheimer disease,” Archives of Neurology , vol. 58, no. 3, pp. 397–405, 2001. [Online]. Available: +http: //dx.doi.org/10.1001/archneur.58.3.397

    work page doi:10.1001/archneur.58.3.397 2001

  3. [3]

    Mapping progressive brain structural changes in early alzheimer’s disease and mild cognitive impairment,

    L. G. Apostolova and P. M. Thompson, “Mapping progressive brain structural changes in early alzheimer’s disease and mild cognitive impairment,” Neuropsychologia, vol. 46, no. 6, pp. 1597 – 1612, 2008, neuroimaging of Early Alzheimer’s Disease. [Online]. Available: http://www.sciencedirect.com/science/article/pii/S0028393207004150

    work page 2008

  4. [4]

    The clinical use of structural MRI in Alzheimer disease,

    G. B. Frisoni, N. C. Fox, C. R. Jack, P. Scheltens, and P. M. Thompson, “The clinical use of structural MRI in Alzheimer disease,” Nat Rev Neurol, vol. 6, no. 2, pp. 67–77, Feb 2010

    work page 2010

  5. [5]

    Refining alzheimer’s disease diagnosis with mri,

    E. Coulthard and M. Knight, “Refining alzheimer’s disease diagnosis with mri,” Brain, vol. 140, no. 3, pp. 524–526, 2017. [Online]. Available: http://dx.doi.org/10.1093/brain/aww335

    work page doi:10.1093/brain/aww335 2017

  6. [6]

    Subjective and objective cognitive decline at the pre- dementia stage of alzheimer’s disease,

    F. Jessen, “Subjective and objective cognitive decline at the pre- dementia stage of alzheimer’s disease,” European Archives of Psychiatry and Clinical Neuroscience, vol. 264, no. 1, pp. 3–7, Nov 2014. [Online]. Available: https://doi.org/10.1007/s00406-014-0539-z

    work page doi:10.1007/s00406-014-0539-z 2014

  7. [7]

    Support vector machine-based classification of Alzheimer’s disease from whole-brain anatomical mri,

    B. Magnin, L. Mesrob, S. Kinkingn ´ehun, M. P ´el´egrini-Issac, O. Colliot, M. Sarazin, B. Dubois, S. Leh ´ericy, and H. Benali, “Support vector machine-based classification of Alzheimer’s disease from whole-brain anatomical mri,” Neuroradiology, vol. 51, no. 2, pp. 73–83, Feb 2009. [Online]. Available: https://doi.org/10.1007/s00234-008-0463-x

    work page doi:10.1007/s00234-008-0463-x 2009

  8. [8]

    Automatic classification of MR scans in Alzheimer’s disease,

    S. Kloppel, C. M. Stonnington, C. Chu, B. Draganski, R. I. Scahill, J. D. Rohrer, N. C. Fox, C. R. Jack, J. Ashburner, and R. S. Frackowiak, “Automatic classification of MR scans in Alzheimer’s disease,” Brain, vol. 131, no. Pt 3, pp. 681–689, Mar 2008

    work page 2008

  9. [9]

    Automatic classification of patients with Alzheimer’s disease from structural MRI: A comparison of ten methods using the ADNI database,

    R. Cuingnet, E. Gerardin, J. Tessieras, G. Auzias, S. Leh ´ericy, M.-O. Habert, M. Chupin, H. Benali, and O. Colliot, “Automatic classification of patients with Alzheimer’s disease from structural MRI: A comparison of ten methods using the ADNI database,” NeuroImage, vol. 56, no. 2, pp. 766 – 781, 2011, multivariate Decoding and Brain Reading. [Online]. Av...

    work page 2011

  10. [10]

    Early prediction of Alzheimer’s Disease with non-local patch-based longitudinal descrip- tors,

    G. Sanroma, V . Andrea, O. M. Benkarim, J. V . Manj ´on, P. Coup ´e, O. Camara, G. Piella, and M. A. Gonz ´alez Ballester, “Early prediction of Alzheimer’s Disease with non-local patch-based longitudinal descrip- tors,” in Patch-Based Techniques in Medical Imaging , G. Wu, B. C. Munsell, Y . Zhan, W. Bai, G. Sanroma, and P. Coup ´e, Eds. Cham: Springer In...

    work page 2017

  11. [11]

    Identification of progressive mild cognitive impairment patients using incomplete longitudinal mri scans,

    K.-H. Thung, C.-Y . Wee, P.-T. Yap, and D. Shen, “Identification of progressive mild cognitive impairment patients using incomplete longitudinal mri scans,” Brain Structure and Function , vol. 221, no. 8, pp. 3979–3995, Nov 2016. [Online]. Available: https: //doi.org/10.1007/s00429-015-1140-6

    work page doi:10.1007/s00429-015-1140-6 2016

  12. [12]

    Early diagnosis of Alzheimer’s Disease by joint feature selection and classification on temporally structured support vector machine,

    Y . Zhu, X. Zhu, M. Kim, D. Shen, and G. Wu, “Early diagnosis of Alzheimer’s Disease by joint feature selection and classification on temporally structured support vector machine,” in Medical Image Com- puting and Computer-Assisted Intervention – MICCAI 2016, S. Ourselin, L. Joskowicz, M. R. Sabuncu, G. Unal, and W. Wells, Eds. Cham: Springer International...

    work page 2016

  13. [13]

    Predicting future clinical changes of mci patients using longitudinal and multimodal biomarkers,

    D. Zhang, D. Shen, and A. D. N. Initiative, “Predicting future clinical changes of mci patients using longitudinal and multimodal biomarkers,” PLOS ONE , vol. 7, no. 3, pp. 1–15, 03 2012. [Online]. Available: https://doi.org/10.1371/journal.pone.0033182

    work page doi:10.1371/journal.pone.0033182 2012

  14. [14]

    The benefits of exercise and metabolic interventions for the prevention and early treatment of Alzheimer’s disease,

    E. Maliszewska-Cyna, M. Lynch, J. Jordan Oore, P. Michael Nagy, and I. Aubert, “The benefits of exercise and metabolic interventions for the prevention and early treatment of Alzheimer’s disease,” Current Alzheimer Research, vol. 14, no. 1, pp. 47–60, 2017. IEEE JOURNAL 11

    work page 2017

  15. [15]

    A Review on Image- and Network-based Brain Data Analysis Techniques for Alzheimer's Disease Diagnosis Reveals a Gap in Developing Predictive Methods for Prognosis

    M. Soussia and I. Rekik, “A review on image-and network-based brain data analysis techniques for Alzheimer’s Disease diagnosis reveals a gap in developing predictive methods for prognosis,” arXiv preprint arXiv:1808.01951, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  16. [16]

    Prediction of longitudinal development of infant cortical surface shape using a 4D current-based learning framework,

    I. Rekik, G. Li, W. Lin, and D. Shen, “Prediction of longitudinal development of infant cortical surface shape using a 4D current-based learning framework,” International Conference on Information Process- ing in Medical Imaging , pp. 576–587, 2015

    work page 2015

  17. [17]

    Predicting infant cortical surface development using a 4D varifold-based learning framework and local topography-based shape morphing,

    ——, “Predicting infant cortical surface development using a 4D varifold-based learning framework and local topography-based shape morphing,” Medical image analysis , vol. 28, pp. 1–12, 2016

    work page 2016

  18. [18]

    Joint prediction of longitudinal development of cortical surfaces and white matter fibers from neonatal MRI,

    I. Rekik, G. Li, P.-T. Yap, G. Chen, W. Lin, and D. Shen, “Joint prediction of longitudinal development of cortical surfaces and white matter fibers from neonatal MRI,” NeuroImage, vol. 152, pp. 411–424, 2017

    work page 2017

  19. [19]

    Learning-based atlas selection for multiple-atlas segmentation,

    G. Sanroma, G. Wu, Y . Gao, and D. Shen, “Learning-based atlas selection for multiple-atlas segmentation,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pp. 3111– 3117, 2014

    work page 2014

  20. [20]

    Visualization and analysis of single-cell rna-seq data by kernel-based similarity learning,

    B. Wang, J. Zhu, E. Pierson, D. Ramazzotti, and S. Batzoglou, “Visualization and analysis of single-cell rna-seq data by kernel-based similarity learning,” Nature Methods, vol. 14, pp. 414 EP –, Mar 2017. [Online]. Available: http://dx.doi.org/10.1038/nmeth.4207

    work page doi:10.1038/nmeth.4207 2017

  21. [21]

    Multiple kernel learning algorithms,

    M. G ¨onen and E. Alpaydın, “Multiple kernel learning algorithms,” Journal of machine learning research , vol. 12, no. Jul, pp. 2211–2268, 2011

    work page 2011

  22. [22]

    The Alzheimer’s disease neuroimaging initiative,

    S. G. Mueller, M. W. Weiner, L. J. Thal, R. C. Petersen, C. Jack, W. Jagust, J. Q. Trojanowski, A. W. Toga, and L. Beckett, “The Alzheimer’s disease neuroimaging initiative,” Neuroimaging Clinics , vol. 15, no. 4, pp. 869–877, 2005

    work page 2005

  23. [23]

    The Alzheimer’s disease neuroimaging initiative (ADNI): MRI meth- ods,

    C. R. Jack Jr, M. A. Bernstein, N. C. Fox, P. Thompson, G. Alexander, D. Harvey, B. Borowski, P. J. Britson, J. L. Whitwell, C. Ward et al. , “The Alzheimer’s disease neuroimaging initiative (ADNI): MRI meth- ods,” Journal of Magnetic Resonance Imaging: An Official Journal of the International Society for Magnetic Resonance in Medicine , vol. 27, no. 4, pp...

    work page 2008

  24. [24]

    Clinical Core of the Alzheimer’s Disease Neuroimaging Initiative: progress and plans,

    P. S. Aisen, R. C. Petersen, M. C. Donohue, A. Gamst, R. Raman, R. G. Thomas, S. Walter, J. Q. Trojanowski, L. M. Shaw, L. A. Beckett et al., “Clinical Core of the Alzheimer’s Disease Neuroimaging Initiative: progress and plans,” Alzheimer’s & Dementia, vol. 6, no. 3, pp. 239–246, 2010

    work page 2010

  25. [25]

    Improved correction for gradient nonlinearity effects in diffusion- weighted imaging,

    E. T. Tan, L. Marinelli, Z. W. Slavens, K. F. King, and C. J. Hardy, “Improved correction for gradient nonlinearity effects in diffusion- weighted imaging,” Journal of Magnetic Resonance Imaging , vol. 38, no. 2, pp. 448–453, 2013

    work page 2013

  26. [26]

    Intensity non-uniformity correction using n3 on 3-t scanners with multichannel phased array coils,

    R. G. Boyes, J. L. Gunter, C. Frost, A. L. Janke, T. Yeatman, D. L. Hill, M. A. Bernstein, P. M. Thompson, M. W. Weiner, N. Schuff et al. , “Intensity non-uniformity correction using n3 on 3-t scanners with multichannel phased array coils,” Neuroimage, vol. 39, no. 4, pp. 1752–1762, 2008

    work page 2008

  27. [27]

    Freesurfer,

    B. Fischl, “Freesurfer,” Neuroimage, vol. 62, no. 2, pp. 774–781, 2012

    work page 2012

  28. [28]

    Automatic 3- D model-based neuroanatomical segmentation,

    D. L. Collins, C. J. Holmes, T. M. Peters, and A. C. Evans, “Automatic 3- D model-based neuroanatomical segmentation,” Human brain mapping , vol. 3, no. 3, pp. 190–208, 1995

    work page 1995

  29. [29]

    Automated 3d mapping of hippocampal atrophy and its clinical correlates in 400 subjects with alzheimer’s disease, mild cognitive impairment, and elderly controls,

    J. H. Morra, Z. Tu, L. G. Apostolova, A. E. Green, C. Avedissian, S. K. Madsen, N. Parikshak, X. Hua, A. W. Toga, C. R. Jack, N. Schuff, M. W. Weiner, and P. M. Thompson, “Automated 3d mapping of hippocampal atrophy and its clinical correlates in 400 subjects with alzheimer’s disease, mild cognitive impairment, and elderly controls,” Human Brain Mapping ,...

    work page doi:10.1002/hbm.20708 2009

  30. [30]

    3d comparison of hippocampal atrophy in amnestic mild cognitive impairment and alzheimer’s disease,

    L. G. Apostolova, I. D. Dinov, R. A. Dutton, K. M. Hayashi, A. W. Toga, J. L. Cummings, and P. M. Thompson, “3d comparison of hippocampal atrophy in amnestic mild cognitive impairment and alzheimer’s disease,” Brain, vol. 129, no. 11, pp. 2867–2873, 2006. [Online]. Available: http://dx.doi.org/10.1093/brain/awl274

    work page doi:10.1093/brain/awl274 2006

  31. [31]

    Memory disorders in probable alzheimer’s disease: the role of hippocampal atrophy as shown with mri

    B. Deweer, S. Leh ´ericy, B. Pillon, M. Baulac, J. Chiras, C. Marsault, Y . Agid, and B. Dubois, “Memory disorders in probable alzheimer’s disease: the role of hippocampal atrophy as shown with mri.” Journal of Neurology, Neurosurgery & Psychiatry , vol. 58, no. 5, pp. 590–597,

    work page

  32. [32]

    Available: http://jnnp.bmj.com/content/58/5/590

    [Online]. Available: http://jnnp.bmj.com/content/58/5/590

    work page

  33. [33]

    The radiologic prediction of alzheimer disease: the atrophic hippocampal formation

    M. J. de Leon, J. Golomb, A. E. George, A. Convit, C. Y . Tarshish, T. McRae, S. De Santi, G. Smith, S. H. Ferris, and M. Noz, “The radiologic prediction of alzheimer disease: the atrophic hippocampal formation.” American Journal of Neuroradiology, vol. 14, no. 4, pp. 897– 906, 1993. [Online]. Available: http://www.ajnr.org/content/14/4/897

    work page 1993

  34. [34]

    Mapping hippocampal and ventricular change in alzheimer disease,

    P. M. Thompson, K. M. Hayashi, G. I. de Zubicaray, A. L. Janke, S. E. Rose, J. Semple, M. S. Hong, D. H. Herman, D. Gravano, D. M. Doddrell, and A. W. Toga, “Mapping hippocampal and ventricular change in alzheimer disease,” NeuroImage, vol. 22, no. 4, pp. 1754 – 1766, 2004. [Online]. Available: http://www.sciencedirect.com/science/ article/pii/S105381190400196X

    work page 2004

  35. [35]

    Ventricular enlargement as a possible measure of alzheimer’s disease progression validated using the alzheimer’s disease neuroimaging initiative database,

    S. M. Nestor, R. Rupsingh, M. Borrie, M. Smith, V . Accomazzi, J. L. Wells, J. Fogarty, R. Bartha, and the Alzheimer’s Disease Neuroimaging Initiative, “Ventricular enlargement as a possible measure of alzheimer’s disease progression validated using the alzheimer’s disease neuroimaging initiative database,” Brain, vol. 131, no. 9, pp. 2443–2454, 2008. [On...

    work page 2008

  36. [36]

    3d maps from multiple mri illustrate changing atrophy patterns as subjects progress from mild cognitive impairment to alzheimer’s disease,

    J. L. Whitwell, S. A. Przybelski, S. D. Weigand, D. S. Knopman, B. F. Boeve, R. C. Petersen, and C. R. Jack, Jr, “3d maps from multiple mri illustrate changing atrophy patterns as subjects progress from mild cognitive impairment to alzheimer’s disease,” Brain, vol. 130, no. 7, pp. 1777–1786, 2007. [Online]. Available: http://dx.doi.org/10.1093/brain/awm112

    work page doi:10.1093/brain/awm112 2007

  37. [37]

    Dynamics of gray matter loss in Alzheimer’s disease,

    P. M. Thompson, K. M. Hayashi, G. de Zubicaray, A. L. Janke, S. E. Rose, J. Semple, D. Herman, M. S. Hong, S. S. Dittmer, D. M. Doddrell, and A. W. Toga, “Dynamics of gray matter loss in Alzheimer’s disease,” J. Neurosci., vol. 23, no. 3, pp. 994–1005, Feb 2003

    work page 2003

  38. [38]

    A comprehensive study of gray matter loss in patients with alzheimer’s disease using optimized voxel-based morphometry,

    G. Karas, E. Burton, S. Rombouts, R. van Schijndel, J. O’Brien, P. Scheltens, I. McKeith, D. Williams, C. Ballard, and F. Barkhof, “A comprehensive study of gray matter loss in patients with alzheimer’s disease using optimized voxel-based morphometry,” NeuroImage, vol. 18, no. 4, pp. 895 – 907, 2003. [Online]. Available: http://www.sciencedirect.com/scien...

    work page 2003

  39. [39]

    Detecting anatomical landmarks for fast Alzheimer’s disease diagnosis,

    J. Zhang, Y . Gao, Y . Gao, B. C. Munsell, and D. Shen, “Detecting anatomical landmarks for fast Alzheimer’s disease diagnosis,” IEEE transactions on medical imaging , vol. 35, no. 12, pp. 2524–2533, 2016

    work page 2016

  40. [40]

    Algorithms for manifold learning,

    L. Cayton, “Algorithms for manifold learning,” Univ. of California at San Diego Tech. Rep , vol. 12, no. 1-17, p. 1, 2005

    work page 2005

  41. [41]

    Riemannian manifold learning for nonlinear dimensionality reduction,

    T. Lin, H. Zha, and S. U. Lee, “Riemannian manifold learning for nonlinear dimensionality reduction,” pp. 44–55, 2006

    work page 2006

  42. [42]

    Magnetic resonance imaging of the entorhinal cortex and hippocampus in mild cognitive impairment and alzheimer’s disease,

    A. T. Du, N. Schuff, D. Amend, M. P. Laakso, Y . Y . Hsu, W. J. Jagust, K. Yaffe, J. H. Kramer, B. Reed, D. Norman, H. C. Chui, and M. W. Weiner, “Magnetic resonance imaging of the entorhinal cortex and hippocampus in mild cognitive impairment and alzheimer’s disease,” Journal of Neurology, Neurosurgery & Psychiatry, vol. 71, no. 4, pp. 441–447, 2001. [On...

    work page 2001

  43. [43]

    The parahippocampal gyrus in alzheimer’s disease: Clinical and preclinical neuroanatomical correlates,

    G. W. V . Hoesen, J. C. Augustinack, J. Dierking, S. J. Redman, and R. Thangavel, “The parahippocampal gyrus in alzheimer’s disease: Clinical and preclinical neuroanatomical correlates,” Annals of the New York Academy of Sciences , vol. 911, no. 1, pp. 254–274, 2006

    work page 2006

  44. [44]

    Geodesic regression of image and shape data for improved modeling of 4d trajectories,

    J. Fishbaugh, M. Prastawa, G. Gerig, and S. Durrleman, “Geodesic regression of image and shape data for improved modeling of 4d trajectories,” Biomedical Imaging (ISBI), 2014 IEEE 11th International Symposium on, pp. 385–388, 2014

    work page 2014

  45. [45]

    Multi-view brain network prediction from a source view using sample selection via cca-based multi-kernel con- nectomic manifold learning,

    M. Zhu and I. Rekik, “Multi-view brain network prediction from a source view using sample selection via cca-based multi-kernel con- nectomic manifold learning,” International Workshop on PRedictive Intelligence In MEdicine , pp. 94–102, 2018

    work page 2018