pith. sign in

arxiv: 2509.09513 · v2 · submitted 2025-09-11 · ⚛️ physics.med-ph · cs.AI· cs.CV· cs.LG· eess.IV

Reduced NEXI protocol for the quantification of human gray matter microstructure on the Connectome 2.0 scanner

Quentin Uhl , Tommaso Pavan , Julianna Gerold , Kwok-Shing Chan , Yohan Jun , Shohei Fujita , Aneri Bhatt , Yixin Ma
show 5 more authors
Qiaochu Wang Hong-Hsi Lee Susie Y. Huang Berkin Bilgic Ileana Jelescu
This is my paper

Pith reviewed 2026-05-18 18:15 UTC · model grok-4.3

classification ⚛️ physics.med-ph cs.AIcs.CVcs.LGeess.IV
keywords Neurite Exchange Imagingdiffusion MRIgray matter microstructureexplainable AIfeature selectionscan time reductionparameter estimationConnectome scanner
0
0 comments X

The pith

An explainable AI framework selects an 8-feature diffusion protocol that halves NEXI acquisition time while matching the theoretical CRLB optimum for gray matter microstructure estimates.

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

The paper develops a shorter protocol for Neurite Exchange Imaging to quantify gray matter microstructure parameters such as compartment diffusivities, neurite fraction, and exchange time. It trains an XGBoost model with SHAP values and recursive feature elimination on synthetic signals to identify a minimal set of eight measurements that cut total scan time from 27 to 14 minutes. In vivo validation in seven healthy participants shows that this reduced protocol reproduces the parameter values and test-retest reliability of the full acquisition. The selected features converge exactly to the Cramér-Rao Lower Bound optimum, demonstrating that the AI approach achieves gold-standard performance without requiring analytical derivatives of the signal model. The method also proves more robust than two simple heuristic sampling schemes that produce biased or noisy estimates.

Core claim

The central claim is that training an explainable AI model on synthetic NEXI signals allows automatic selection of an optimal reduced acquisition protocol whose eight features suffice to recover the same microstructural parameter estimates as the original fifteen-feature protocol, match the Cramér-Rao Lower Bound performance bound, and maintain in vivo test-retest reproducibility on the Connectome 2.0 scanner.

What carries the argument

The XAI pipeline that combines XGBoost regression, SHAP importance scores, and recursive feature elimination to rank and prune diffusion measurement features from synthetic data generated by the NEXI model.

If this is right

  • The 14-minute protocol supports faster exchange-sensitive microstructural mapping in research and clinical settings.
  • XAI feature selection can replace CRLB optimization for any numerical biophysical model where analytical Jacobians are intractable.
  • Heuristic sampling schemes such as mid-range or corner placement produce systematically biased or unstable parameter maps.
  • Test-retest reproducibility of NEXI parameters remains intact after reduction to the XAI-selected subset.

Where Pith is reading between the lines

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

  • The same XAI procedure could be retrained on data from conventional clinical scanners to derive reduced protocols without ultra-high gradients.
  • Applying the protocol to patient groups with altered gray-matter exchange properties would test whether the reduced set remains sensitive to pathology.
  • Because the XAI selection converged to the CRLB optimum, the approach may generalize to other multi-compartment diffusion models that lack closed-form derivatives.

Load-bearing premise

Synthetic signals generated from the NEXI model capture the noise characteristics and signal behavior of real in vivo human brain data acquired on the Connectome 2.0 scanner.

What would settle it

In a new cohort of in vivo scans, the XAI-selected eight-feature protocol yields exchange-time or intra-neurite diffusivity estimates whose coefficient of variation exceeds that of the full protocol by more than a factor of two.

Figures

Figures reproduced from arXiv: 2509.09513 by Aneri Bhatt, Berkin Bilgic, Hong-Hsi Lee, Ileana Jelescu, Julianna Gerold, Kwok-Shing Chan, Qiaochu Wang, Quentin Uhl, Shohei Fujita, Susie Y. Huang, Tommaso Pavan, Yixin Ma, Yohan Jun.

Figure 1
Figure 1. Figure 1: Cortical surface maps of NEXI parameters from the full protocol. Group-averaged maps of tex, f, Di, and De across all subjects, displayed on lateral and medial views of both cortical hemispheres. As expected, tex and f are elevated in the motor cortex, consistent with prior associations to increased myelination and neurite density. However, unlike earlier reports on Connectome 1.0 and Prisma systems, no cl… view at source ↗
Figure 4
Figure 4. Figure 4: Distributions and summary of NEXI parameter estimates across DKT cortical regions. (A) Kernel density estimates of voxel￾level values for each NEXI parameter (tex, f, De, Di), aggregated across all DKT ROIs and all subjects. The full protocol (top row) and XAI-reduced protocol (bottom row) yield very similar distributions. The vertical dashed lines indicate the mode of each distribution, reflecting the mos… view at source ↗
Figure 3
Figure 3. Figure 3: Estimation error across the parameter space: Full vs. XAI￾optimized protocol. Binned error analysis from 100,000 synthetic NEXI signals with added Gaussian noise (SNR = 32 ± 26), comparing the full protocol (blue) and the reduced 8-feature XAI protocol (orange). For each NEXI parameter (tex, f, Di, De), absolute estimation errors are plotted as a function of the ground-truth parameter value. Histograms abo… view at source ↗
Figure 6
Figure 6. Figure 6: Root Mean Square Deviation (RMSD) and estimation bias of NEXI parameters across protocols. Each point corresponds to the RMSD between the median value from the full protocol and that from a reduced subprotocol (XAI-optimized, FIM, or naive) within a specific DKT ROI for a given subject, aggregated across all ROIs and subjects. RMSD and median bias values for all NEXI parameters (tex, f, De, Di) are present… view at source ↗
Figure 5
Figure 5. Figure 5: Comparison of reduced protocols: FIM-optimized vs. naive. (A) Schematic of the Fisher Information Matrix (FIM)-based Recursive Feature Elimination (FIM-RFE) algorithm. At each step, all possible subprotocols with one fewer feature are evaluated, and the one yielding the highest log det(FIM) is retained. The process is repeated until only a single feature remains, and the curve of log det(FIM) vs. number of… view at source ↗
Figure 7
Figure 7. Figure 7: Left hemisphere cortical maps of NEXI parameters across all acquisition protocols. Group-averaged cortical surface maps for tex, f, De, and Di are shown for four protocols: Full, XAI-optimized (SHAP￾based), FIM-optimized, and naive. While all protocols capture general spatial trends, the XAI protocol best reproduces the contrasts seen in the full protocol, particularly elevated tex and f in sensorimotor an… view at source ↗
Figure 8
Figure 8. Figure 8: Test-retest reproducibility of NEXI estimates using the full protocol and the XAI-optimized subprotocol. Bland-Altman plots show intra-subject variability across scan-rescan sessions, using DKT ROI median values for each NEXI parameter (tex, f, Di, De). Left: Full protocol. Right: XAI-Reduced protocol. Each plot displays the difference between sessions against their mean, with orange lines indicating bias … view at source ↗
read the original abstract

Biophysical diffusion MRI models like Neurite Exchange Imaging (NEXI) are essential for probing gray matter microstructure, estimating compartment diffusivities, neurite fraction, and exchange time. However, NEXI's multi-shell, multi-diffusion-time requirements cause prohibitively long acquisitions. Leveraging the Connectome 2.0 ultra-high gradient scanner, we developed a time-efficient protocol using an Explainable AI (XAI) framework. Combining XGBoost, SHAP, and Recursive Feature Elimination trained on synthetic signals, XAI identified an optimal 8-feature subset, cutting scan time from 27 to 14 minutes. Validated in vivo in seven healthy participants, the XAI protocol was benchmarked against the full 15-feature acquisition, a Cram'er-Rao Lower Bound (CRLB) theoretical optimum, and two heuristics ("Mid-Range" and "Corner"). It robustly reproduced parameter estimates and maintained test-retest reproducibility. Remarkably, the XAI selection converged to the CRLB optimum. This validates XAI's optimality while highlighting its main advantage: achieving gold-standard optimization without complex analytical Jacobians, making it easily adaptable to numerical models or complex noise where CRLB is intractable. Furthermore, XAI showed superior in vivo robustness over heuristics: "Mid-Range" sampling yielded biased exchange time estimates from insufficient temporal diversity, while "Corner" sampling gave unstable intra-neurite diffusivity estimates (5-fold higher CV) due to noise sensitivity. Ultimately, this robust 14-minute protocol accelerates exchange-sensitive microstructural mapping, establishing a model-agnostic optimization framework adaptable to future ultra-high gradient systems and existing clinical scanners.

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

2 major / 2 minor

Summary. The paper develops a reduced 8-feature NEXI protocol for gray matter microstructure on the Connectome 2.0 scanner by applying an XAI pipeline (XGBoost + SHAP + RFE) to synthetic signals generated from the NEXI forward model. The selected protocol shortens acquisition from 27 to 14 minutes and is validated in vivo on seven healthy participants, where it is shown to reproduce full-protocol parameter estimates, maintain test-retest reproducibility, converge to the CRLB optimum, and outperform two heuristic selections (Mid-Range and Corner) in bias and coefficient-of-variation metrics.

Significance. If the central claims hold, the work supplies a practical, model-agnostic route to protocol optimization for exchange-sensitive diffusion models that avoids the analytical Jacobians required by classical CRLB design. The explicit in-vivo comparison to both the theoretical optimum and simple heuristics, together with the reported test-retest stability, constitutes a concrete advance for ultra-high-gradient microstructural imaging. The approach could be extended to other numerical or complex-noise models where CRLB is intractable.

major comments (2)
  1. [Abstract / Results] Abstract and Results (in-vivo validation paragraph): the claim that the XAI protocol 'robustly reproduced parameter estimates' rests on a cohort of only seven participants; no quantitative agreement metrics (e.g., mean absolute percentage error, ICC, or Bland-Altman limits) between the 8-feature and 15-feature protocols are supplied, nor are error bars or statistical tests for the reported superiority over heuristics. This weakens the robustness conclusion that is central to the paper's practical utility.
  2. [Methods] Methods (XAI training and feature-selection subsection): feature selection is performed exclusively on synthetic signals generated from the NEXI model under idealized noise assumptions. No sensitivity analysis to Connectome 2.0-specific effects (gradient nonlinearities, physiological noise, motion, or partial-volume) is presented, yet the in-vivo results with n=7 are used to assert optimality and convergence to CRLB. Because the selected features are load-bearing for all subsequent claims, this gap requires either additional real-data ablation or explicit justification that the synthetic distribution suffices.
minor comments (2)
  1. [Abstract] The abstract states that XAI 'achieves gold-standard optimization without complex analytical Jacobians,' but does not clarify whether the CRLB reference itself was computed numerically or analytically; a brief statement in the Methods would remove ambiguity.
  2. [Results] Table or figure reporting test-retest CVs should include the number of repeated scans per subject and the exact definition of CV (within-subject or pooled) to allow direct comparison with literature values.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which highlight important aspects of our validation and methods. We address each major comment below and indicate where revisions will be made to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract / Results] Abstract and Results (in-vivo validation paragraph): the claim that the XAI protocol 'robustly reproduced parameter estimates' rests on a cohort of only seven participants; no quantitative agreement metrics (e.g., mean absolute percentage error, ICC, or Bland-Altman limits) between the 8-feature and 15-feature protocols are supplied, nor are error bars or statistical tests for the reported superiority over heuristics. This weakens the robustness conclusion that is central to the paper's practical utility.

    Authors: We agree that the in-vivo cohort size of seven participants is modest and that additional quantitative metrics would strengthen the claims of reproduction and superiority. The current manuscript reports test-retest reproducibility and qualitative reproduction of parameter estimates, along with bias and coefficient-of-variation comparisons to heuristics. In the revised version we will add mean absolute percentage error, intraclass correlation coefficients, and Bland-Altman limits of agreement between the 8-feature and 15-feature protocols. We will also include error bars on the relevant figures and report the results of paired statistical tests (e.g., t-tests or Wilcoxon signed-rank) for differences versus the heuristic protocols, with p-values provided. revision: yes

  2. Referee: [Methods] Methods (XAI training and feature-selection subsection): feature selection is performed exclusively on synthetic signals generated from the NEXI model under idealized noise assumptions. No sensitivity analysis to Connectome 2.0-specific effects (gradient nonlinearities, physiological noise, motion, or partial-volume) is presented, yet the in-vivo results with n=7 are used to assert optimality and convergence to CRLB. Because the selected features are load-bearing for all subsequent claims, this gap requires either additional real-data ablation or explicit justification that the synthetic distribution suffices.

    Authors: The XAI feature selection is intentionally model-based, using synthetic signals generated from the NEXI forward model over physiologically plausible parameter ranges drawn from the literature. This ensures the selected features are optimal for recovering the model parameters under the assumed signal and noise model. The in-vivo experiments then serve as an independent validation step, showing that the chosen protocol reproduces full-protocol estimates, maintains test-retest stability, and converges to the CRLB optimum. We will add an explicit justification in the revised Methods and Discussion sections explaining why the synthetic distribution is appropriate for this purpose, while acknowledging scanner-specific effects as a limitation. A brief sensitivity analysis exploring added physiological noise or partial-volume effects will also be included if it can be performed with the existing simulation framework. revision: partial

Circularity Check

0 steps flagged

No significant circularity detected in the protocol optimization chain

full rationale

The paper trains XGBoost/SHAP/RFE on synthetic signals generated from the NEXI forward model to identify an 8-feature subset, then validates the resulting protocol on real in vivo data from seven participants by direct comparison to the full 15-feature acquisition, the CRLB theoretical optimum, and two heuristic protocols, while also reporting test-retest reproducibility. This chain does not reduce to its inputs by construction: the synthetic training step is a standard simulation-based optimization whose claims are grounded in independent empirical measurements on actual Connectome 2.0 scans rather than tautological re-derivation of the same synthetic quantities. No self-definitional equations, fitted-input predictions, or load-bearing self-citations appear in the derivation.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim depends on the NEXI biophysical model being an adequate description of gray matter signals and on synthetic data being representative of real scanner noise and tissue properties.

free parameters (1)
  • NEXI compartment parameters (intra-neurite diffusivity, extra-neurite diffusivity, neurite fraction, exchange time)
    These are used to generate the synthetic training signals on which the XAI feature selection is performed.
axioms (1)
  • domain assumption Synthetic signals generated from the NEXI model accurately mimic in vivo diffusion MRI signals including noise on the Connectome 2.0 scanner
    Invoked when training XGBoost and applying SHAP/RFE to select the 8-feature subset.

pith-pipeline@v0.9.0 · 5897 in / 1466 out tokens · 54599 ms · 2026-05-18T18:15:44.667914+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

53 extracted references · 53 canonical work pages

  1. [1]

    Diffusion MRI biomarkers of white matter microstructure vary nonmono- tonically with increasing cerebral amyloid deposition,

    J. W. Dong, I. O. Jelescu, B. Ades-Aron, D. S. Novikov, K. Friedman, J. S. Babb, R. S. Osorio, J. E. Galvin, T. M. Shepherd, and E. Fieremans, “Diffusion MRI biomarkers of white matter microstructure vary nonmono- tonically with increasing cerebral amyloid deposition,”Neurobiology of Aging, vol. 89, pp. 118–128, May 2020

    work page 2020

  2. [2]

    Diffusion tensor imaging in Parkinson’s disease: Review and meta-analysis,

    C. Atkinson-Clement, S. Pinto, A. Eusebio, and O. Coulon, “Diffusion tensor imaging in Parkinson’s disease: Review and meta-analysis,” NeuroImage: Clinical, vol. 16, pp. 98–110, Jan. 2017

    work page 2017

  3. [3]

    Measures of cortical microstructure are linked to amyloid pathology in Alzheimer’s disease,

    N. Spotorno, O. Strandberg, G. Vis, E. Stomrud, M. Nilsson, and O. Hansson, “Measures of cortical microstructure are linked to amyloid pathology in Alzheimer’s disease,”Brain, vol. 146, no. 4, pp. 1602–1614, Apr. 2023

    work page 2023

  4. [4]

    Estimation of the Effective Self-DiffusionTensorfrom the NMR Spin Echo,

    P. J. Basser, J. Mattiello, and D. Lebihan, “Estimation of the Effective Self-DiffusionTensorfrom the NMR Spin Echo,”Journal of Magnetic Resonance, Series B, vol. 103, no. 3, pp. 247–254, Mar. 1994

    work page 1994

  5. [5]

    Imaging of diffusion and microcirculation with gradient sensitization: Design, strategy, and significance,

    D. Le Bihan, R. Turner, C. T. W. Moonen, and J. Pekar, “Imaging of diffusion and microcirculation with gradient sensitization: Design, strategy, and significance,”Journal of Magnetic Resonance Imaging, vol. 1, no. 1, pp. 7–28, Jan. 1991

    work page 1991

  6. [6]

    Imaging brain microstructure with diffusion MRI: Practicality and applications,

    D. C. Alexander, T. B. Dyrby, M. Nilsson, and H. Zhang, “Imaging brain microstructure with diffusion MRI: Practicality and applications,”NMR in Biomedicine, vol. 32, no. 4, p. e3841, 2019, e3841 NBM-16-0261.R1

    work page 2019

  7. [7]

    Challenges for biophysical modeling of microstructure,

    I. O. Jelescu, M. Palombo, F. Bagnato, and K. G. Schilling, “Challenges for biophysical modeling of microstructure,”Journal of Neuroscience Methods, vol. 344, p. 108861, Oct. 2020

    work page 2020

  8. [8]

    Quantifying brain microstructure with diffusion MRI: Theory and parameter estimation,

    D. S. Novikov, E. Fieremans, S. N. Jespersen, and V . G. Kiselev, “Quantifying brain microstructure with diffusion MRI: Theory and parameter estimation,”NMR in Biomedicine, vol. 32, no. 4, p. e3998, 2019

    work page 2019

  9. [9]

    An analytical model of restricted diffusion in bovine optic nerve,

    G. J. Stanisz, G. A. Wright, R. M. Henkelman, and A. Szafer, “An analytical model of restricted diffusion in bovine optic nerve,”Magnetic Resonance in Medicine, vol. 37, no. 1, pp. 103–111, 1997

    work page 1997

  10. [10]

    Neurite Exchange Imaging (NEXI): A minimal model of diffusion in gray matter with inter-compartment water exchange,

    I. O. Jelescu, A. de Skowronski, F. Geffroy, M. Palombo, and D. S. Novikov, “Neurite Exchange Imaging (NEXI): A minimal model of diffusion in gray matter with inter-compartment water exchange,” NeuroImage, vol. 256, p. 119277, Aug. 2022

    work page 2022

  11. [11]

    SANDI: A compartment-based model for non-invasive apparent soma and neurite imaging by diffusion MRI,

    M. Palombo, A. Ianus, M. Guerreri, D. Nunes, D. C. Alexander, N. Shemesh, and H. Zhang, “SANDI: A compartment-based model for non-invasive apparent soma and neurite imaging by diffusion MRI,” NeuroImage, vol. 215, p. 116835, Jul. 2020

    work page 2020

  12. [12]

    In vivo observation and biophysical interpretation of time-dependent diffusion in human cortical gray matter,

    H.-H. Lee, A. Papaioannou, D. S. Novikov, and E. Fieremans, “In vivo observation and biophysical interpretation of time-dependent diffusion in human cortical gray matter,”NeuroImage, vol. 222, p. 117054, Nov. 2020

    work page 2020

  13. [13]

    Diffusion time dependence, power-law scaling, and exchange in gray matter,

    J. L. Olesen, L. Østergaard, N. Shemesh, and S. N. Jespersen, “Diffusion time dependence, power-law scaling, and exchange in gray matter,” NeuroImage, vol. 251, p. 118976, May 2022

    work page 2022

  14. [14]

    Quantifying human gray matter microstructure using neurite exchange imaging (NEXI) and 300 mT/m gradients,

    Q. Uhl, T. Pavan, M. Molendowska, D. K. Jones, M. Palombo, and I. O. Jelescu, “Quantifying human gray matter microstructure using neurite exchange imaging (NEXI) and 300 mT/m gradients,”Imaging Neuroscience, vol. 2, pp. 1–19, Mar. 2024

    work page 2024

  15. [15]

    Human gray matter microstructure mapped using neurite exchange imaging (NEXI) on a clinical scanner,

    Q. Uhl, T. Pavan, T. Feiweier, G. F. Piredda, and I. Jelescu, “Human gray matter microstructure mapped using neurite exchange imaging (NEXI) on a clinical scanner,”Imaging Neuroscience, vol. 3, p. IMAG.a.32, Jun. 2025

    work page 2025

  16. [16]

    Cortical microstructure in primary progressive aphasia: A multicenter study,

    I. Ill´an-Gala, V . Montal, S. Borrego-´Ecija, M. L. Mandelli, N. Falg`as, A. E. Welch, J. Pegueroles, M. Santos-Santos, A. Bejanin, D. Alcolea, O. Dols- Icardo, O. Belbin, M. B. S ´anchez-Saudin´os, N. Bargall ´o, S. Gonz ´alez- Ortiz, A. Llad ´o, R. Blesa, B. C. Dickerson, H. J. Rosen, B. L. Miller, A. Lle ´o, M. L. Gorno-Tempini, R. S ´anchez-Valle, an...

    work page 2022

  17. [17]

    Connectome 2.0: Developing the next-generation ultra-high gradient strength human MRI scanner for bridging studies of the micro-, meso- and macro-connectome,

    S. Y . Huang, T. Witzel, B. Keil, A. Scholz, M. Davids, P. Dietz, E. Rummert, R. Ramb, J. E. Kirsch, A. Yendiki, Q. Fan, Q. Tian, G. Ramos-Llord´en, H.-H. Lee, A. Nummenmaa, B. Bilgic, K. Setsompop, F. Wang, A. V . Avram, M. Komlosh, D. Benjamini, K. N. Magdoom, S. Pathak, W. Schneider, D. S. Novikov, E. Fieremans, S. Tounekti, C. Mekkaoui, J. Augustinack...

    work page 2021

  18. [18]

    Ultra-high gradient connectomics and microstructure MRI scanner for imaging of human brain circuits across scales,

    G. Ramos-Llord ´en, H.-H. Lee, M. Davids, P. Dietz, A. Krug, J. E. Kirsch, M. Mahmutovic, A. M ¨uller, Y . Ma, H. Lee, C. Maffei, A. Yendiki, B. Bilgic, D. J. Park, Q. Tian, B. Clifford, W.-C. Lo, S. Stocker, J. Fischer, G. Ruyters, M. Roesler, A. Potthast, T. Benner, E. Rummert, R. Schuster, P. J. Basser, T. Witzel, L. L. Wald, B. R. Rosen, B. Keil, and ...

    work page 2025

  19. [19]

    In vivo human neurite exchange time imaging at 500 mT/m diffusion gradients,

    K.-S. Chan, Y . Ma, H. Lee, J. P. Marques, J. L. Olesen, S. Coelho, D. S. Novikov, S. N. Jespersen, S. Y . Huang, and H.-H. Lee, “In vivo human neurite exchange time imaging at 500 mT/m diffusion gradients,” Imaging Neuroscience, Apr. 2025

    work page 2025

  20. [20]

    Optimization of b-value schemes for estimation of the diffusion coefficient and the perfusion fraction with segmented intravoxel incoherent motion model fitting,

    O. Jalnefjord, M. Montelius, G. Starck, and M. Ljungberg, “Optimization of b-value schemes for estimation of the diffusion coefficient and the perfusion fraction with segmented intravoxel incoherent motion model fitting,”Magnetic Resonance in Medicine, vol. 82, no. 4, pp. 1541–1552, Oct. 2019

    work page 2019

  21. [21]

    Robust design of biological experiments,

    P. Flaherty, A. Arkin, and M. Jordan, “Robust design of biological experiments,” inAdvances in Neural Information Processing Systems, vol. 18. MIT Press, 2005

    work page 2005

  22. [22]

    XGBoost: A scalable tree boosting system,

    T. Chen and C. Guestrin, “XGBoost: A scalable tree boosting system,” inProceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ser. Kdd ’16. San Francisco, California, USA and New York, NY , USA: Association for Computing Machinery, 2016, pp. 785–794

    work page 2016

  23. [23]

    A unified approach to interpreting model predictions,

    S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” inAdvances in Neural Information Processing Systems, I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, Eds., vol. 30. Curran Associates, Inc., 2017, pp. 4768–4777. 10

    work page 2017

  24. [24]

    An integrated approach of feature selection and machine learning for early detection of breast cancer,

    J. Zhu, Z. Zhao, C. Wu, C. Yin, R. Chen, and Y . Ding, “An integrated approach of feature selection and machine learning for early detection of breast cancer,”Scientific Reports, vol. 15, no. 1, p. 13015, Apr. 2025

    work page 2025

  25. [25]

    Performance and explainability of feature selection-boosted tree-based classifiers for COVID-19 detection,

    J. Rufino, J. M. Ram ´ırez, J. Aguilar, C. Baquero, J. Champati, D. Frey, R. E. Lillo, and A. Fern ´andez-Anta, “Performance and explainability of feature selection-boosted tree-based classifiers for COVID-19 detection,” Heliyon, vol. 10, no. 1, p. e23219, Jan. 2024

    work page 2024

  26. [26]

    A multilayer stacking method base on RFE-SHAP feature selection strategy for recognition of driver’s mental load and emotional state,

    J. Huang, Y . Peng, and L. Hu, “A multilayer stacking method base on RFE-SHAP feature selection strategy for recognition of driver’s mental load and emotional state,”Expert Systems with Applications, vol. 238, p. 121729, Mar. 2024

    work page 2024

  27. [27]

    Coupling Interpretable Feature Selection with Machine Learning for Evapotranspiration Gap Filling,

    L. Wang, L. Dong, and Q. Zhang, “Coupling Interpretable Feature Selection with Machine Learning for Evapotranspiration Gap Filling,” Water, vol. 17, no. 5, p. 748, Jan. 2025

    work page 2025

  28. [28]

    Hybrid bagging and boosting with SHAP based feature selection for enhanced predictive modeling in intrusion detection systems,

    U. Ahmed, Z. Jiangbin, A. Almogren, M. Sadiq, A. U. Rehman, M. T. Sadiq, and J. Choi, “Hybrid bagging and boosting with SHAP based feature selection for enhanced predictive modeling in intrusion detection systems,”Scientific Reports, vol. 14, no. 1, p. 30532, Dec. 2024

    work page 2024

  29. [29]

    Prediction of Customer Satisfaction Using RFE-SHAP Feature Selection Method,

    O. Chernyaeva and T. Hong, “Prediction of Customer Satisfaction Using RFE-SHAP Feature Selection Method,”Journal of Intelligence and Information Systems, vol. 29, no. 4, pp. 325–345, 2023

    work page 2023

  30. [30]

    High-density MRI coil arrays with integrated field monitoring systems for human connectome mapping,

    M. Mahmutovic, M. Shrestha, G. Ramos-Llord´en, D. Sung, L. J. Edwards, Y . Chu, P. I. Dubovan, A. M¨uller, S.-L. J. Hansen, A. Ghotra, K. J. Pine, R. M ¨uller, N. Weiskopf, L. L. Wald, C. Mekkaoui, H. E. M ¨oller, S. Y . Huang, and B. Keil, “High-density MRI coil arrays with integrated field monitoring systems for human connectome mapping,”Magnetic Resona...

    work page 2025

  31. [31]

    Denoising of diffusion MRI using random matrix theory,

    J. Veraart, D. S. Novikov, D. Christiaens, B. Ades-aron, J. Sijbers, and E. Fieremans, “Denoising of diffusion MRI using random matrix theory,” NeuroImage, vol. 142, pp. 394–406, Nov. 2016

    work page 2016

  32. [32]

    Gibbs-ringing artifact removal based on local subvoxel-shifts,

    E. Kellner, B. Dhital, V . G. Kiselev, and M. Reisert, “Gibbs-ringing artifact removal based on local subvoxel-shifts,”Magnetic Resonance in Medicine, vol. 76, no. 5, pp. 1574–1581, Nov. 2016

    work page 2016

  33. [33]

    An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging,

    J. L. R. Andersson and S. N. Sotiropoulos, “An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging,”NeuroImage, vol. 125, pp. 1063–1078, Jan. 2016

    work page 2016

  34. [34]

    FastSurfer - A fast and accurate deep learning based neuroimaging pipeline,

    L. Henschel, S. Conjeti, S. Estrada, K. Diers, B. Fischl, and M. Reuter, “FastSurfer - A fast and accurate deep learning based neuroimaging pipeline,”NeuroImage, vol. 219, p. 117012, Oct. 2020

    work page 2020

  35. [35]

    101 Labeled Brain Images and a Consistent Human Cortical Labeling Protocol,

    A. Klein and J. Tourville, “101 Labeled Brain Images and a Consistent Human Cortical Labeling Protocol,”Frontiers in Neuroscience, vol. 6, 2012

    work page 2012

  36. [36]

    Advanced Normalization Tools: V1.0,

    B. Avants, N. J. Tustison, and G. Song, “Advanced Normalization Tools: V1.0,”The Insight Journal, Jul. 2009

    work page 2009

  37. [37]

    Informatics and Data Mining Tools and Strategies for the Human Connectome Project,

    D. Marcus, J. Harwell, T. Olsen, M. Hodge, M. Glasser, F. Prior, M. Jenkinson, T. Laumann, S. Curtiss, and D. Van Essen, “Informatics and Data Mining Tools and Strategies for the Human Connectome Project,” Frontiers in Neuroinformatics, vol. 5, 2011

    work page 2011

  38. [38]

    Monte Carlo study of a two-compartment exchange model of diffusion,

    E. Fieremans, D. S. Novikov, J. H. Jensen, and J. A. Helpern, “Monte Carlo study of a two-compartment exchange model of diffusion,”NMR in Biomedicine, vol. 23, no. 7, pp. 711–724, 2010

    work page 2010

  39. [39]

    Diffusion-time dependence of diffusional kurtosis in the mouse brain,

    M. Aggarwal, M. D. Smith, and P. A. Calabresi, “Diffusion-time dependence of diffusional kurtosis in the mouse brain,”Magnetic Resonance in Medicine, vol. 84, no. 3, pp. 1564–1578, 2020

    work page 2020

  40. [40]

    Time-dependent diffusivity and kurtosis in phantoms and patients with head and neck cancer,

    E. Solomon, G. Lemberskiy, S. Baete, K. Hu, D. Malyarenko, S. Swanson, A. Shukla-Dave, S. E. Russek, E. Zan, and S. G. Kim, “Time-dependent diffusivity and kurtosis in phantoms and patients with head and neck cancer,”Magnetic Resonance in Medicine, vol. 89, no. 2, pp. 522–535, Feb. 2023

    work page 2023

  41. [41]

    Time- Dependent Diffusion & Kurtosis as a Probe of Tissue Structure,

    D. S. Novikov, E. Fieremans, J. H. Jensen, and J. A. Helpern, “Time- Dependent Diffusion & Kurtosis as a Probe of Tissue Structure,” inProc. Intl. Soc. Mag. Reson. Med. 2011, Montreal, 2011, p. 1988

    work page 2011

  42. [42]

    Parameter estimation from Rician-distributed data sets using a maximum likelihood estimator: Application to t1 and perfusion measurements,

    O. T. Karlsen, R. Verhagen, and W. M. Bov ´ee, “Parameter estimation from Rician-distributed data sets using a maximum likelihood estimator: Application to t1 and perfusion measurements,”Magnetic Resonance in Medicine, vol. 41, no. 3, pp. 614–623, 1999

    work page 1999

  43. [43]

    Microscopic anisotropy misestimation in spherical-mean single diffusion encoding MRI,

    R. N. Henriques, S. N. Jespersen, and N. Shemesh, “Microscopic anisotropy misestimation in spherical-mean single diffusion encoding MRI,”Magnetic resonance in medicine, vol. 81, no. 5, pp. 3245–3261, 2019

    work page 2019

  44. [44]

    GEM: A unifying model for Gray Matter microstructure,

    Q. Uhl, T. Pavan, I. de Riedmatten, J. Nguyen-Duc, and I. Jelescu, “GEM: A unifying model for Gray Matter microstructure,” inProc. Intl. Soc. Mag. Reson. Med. 2024, Singapore, Singapore, 2024, p. 3455

    work page 2024

  45. [45]

    P. F. M. Ribeiro, L. Ventura-Antunes, M. Gabi, B. Mota, L. T. Grinberg, J. M. Farfel, R. E. L. Ferretti-Rebustini, R. E. P. Leite, W. J. Filho, and S. Herculano-Houzel, “The human cerebral cortex is neither one nor many: Neuronal distribution reveals two quantitatively different zones in the gray matter, three in the white matter, and explains local varia...

    work page 2013

  46. [46]

    Degeneracy in model parameter estimation for multi-compartmental diffusion in neuronal tissue: Degeneracy in Model Parameter Estimation of Diffusion in Neural Tissue,

    I. O. Jelescu, J. Veraart, E. Fieremans, and D. S. Novikov, “Degeneracy in model parameter estimation for multi-compartmental diffusion in neuronal tissue: Degeneracy in Model Parameter Estimation of Diffusion in Neural Tissue,”NMR in Biomedicine, vol. 29, no. 1, pp. 33–47, Jan. 2016

    work page 2016

  47. [47]

    Estimating axial diffusivity in the NODDI model,

    A. F. Howard, M. Cottaar, M. Drakesmith, Q. Fan, S. Y . Huang, D. K. Jones, F. J. Lange, J. Mollink, S. U. Rudrapatna, Q. Tian, K. L. Miller, and S. Jbabdi, “Estimating axial diffusivity in the NODDI model,” NeuroImage, vol. 262, p. 119535, Nov. 2022

    work page 2022

  48. [48]

    Exchange between structurally-disordered compartments,

    D. S. Novikov, R. Coronado-Leija, and E. Fieremans, “Exchange between structurally-disordered compartments,” inProc. Intl. Soc. Mag. Reson. Med. 2023, Toronto, Canada, 2023, p. 0684

    work page 2023

  49. [49]

    Intra-axonal diffusivity in brain white matter,

    B. Dhital, M. Reisert, E. Kellner, and V . G. Kiselev, “Intra-axonal diffusivity in brain white matter,”NeuroImage, vol. 189, pp. 543–550, Apr. 2019

    work page 2019

  50. [50]

    Efficient PCA denoising of spatially correlated redundant MRI data,

    R. N. Henriques, A. Ianu s ¸, L. Novello, J. Jovicich, S. N. Jespersen, and N. Shemesh, “Efficient PCA denoising of spatially correlated redundant MRI data,”Imaging Neuroscience, vol. 1, pp. 1–26, Dec. 2023

    work page 2023

  51. [51]

    NOise reduction with DIstribution Corrected (NORDIC) PCA in dMRI with complex-valued parameter-free locally low-rank processing,

    S. Moeller, P. K. Pisharady, S. Ramanna, C. Lenglet, X. Wu, L. Dowdle, E. Yacoub, K. U ˘gurbil, and M. Ak c ¸akaya, “NOise reduction with DIstribution Corrected (NORDIC) PCA in dMRI with complex-valued parameter-free locally low-rank processing,”NeuroImage, vol. 226, p. 117539, Feb. 2021

    work page 2021

  52. [52]

    Optimizing the NEXI acquisition protocol for quantifying human gray matter microstructure on a clinical MRI scanner using Explainable AI,

    Q. Uhl, T. Pavan, T. Feiweier, E. J. Canales-Rodr ´ıguez, and I. O. Jelescu, “Optimizing the NEXI acquisition protocol for quantifying human gray matter microstructure on a clinical MRI scanner using Explainable AI,” inProc. Intl. Soc. Mag. Reson. Med. 2023, Toronto, Canada, Jun. 2023, p. 0948

    work page 2023

  53. [53]

    Diffusion MRI with free gradient waveforms on a high- performance gradient system: Probing restriction and exchange in the human brain,

    A. Chakwizira, A. Zhu, T. Foo, C.-F. Westin, F. Szczepankiewicz, and M. Nilsson, “Diffusion MRI with free gradient waveforms on a high- performance gradient system: Probing restriction and exchange in the human brain,”NeuroImage, vol. 283, p. 120409, Dec. 2023. Explainable AI for Accelerated Microstructure Imaging: A SHAP-Guided Protocol on the Connectome...

    work page 2023