MACAW: A Causal Generative Model for Medical Imaging

Erik Ohara; Matthias Wilms; Nils Forkert; Vibujithan Vigneshwaran

REVIEW 4 major objections 5 minor 1 cited by

MACAW: A Causal Generative Model for Medical Imaging

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a single normalizing flow whose autoregressive conditioners are masked by a causal DAG's adjacency matrix can answer associational, interventional, and counterfactual queries on high-dimensional medical images, and…

desk verdict The synthetic experiment gives the method real teeth, but the UK Biobank results rest on an untested factorization that the paper never names as a limitation. read the letter →

arxiv 2412.02900 v1 pith:M7NQMSYG submitted 2024-12-03 eess.IV cs.LG

Vibujithan Vigneshwaran , Erik Ohara , Matthias Wilms , Nils Forkert This is my paper · ORCID

classification eess.IVcs.LG

keywords causalgenerativemodelnormalizingflowscounterfactualinferencemedicalimagingneuroimagingbrainagingBayesianclassificationmaskedautoencoders

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that one invertible generative model, called Masked Causal Flow (MACAW), can encode a full causal structure into a medical image generator. Prior causal image models required a separate normalizing flow per causal variable, which limited them to a few variables. MACAW instead masks the connections of a single flow's neural-network conditioners according to the causal graph, so the model's density factorization matches the causal factorization. If it works, the same trained model can draw new brain images, simulate interventions such as setting age to 75, generate individual-level counterfactual images of aging, and produce Bayesian age predictions with uncertainty. The authors test it on a synthetic structural equation model and on 23,692 MRI slices from a large population cohort, where the counterfactual images show ventricular enlargement with age in line with known brain-aging anatomy.

What carries the argument

The load-bearing object is the causally-masked autoencoder (C-MADE), an autoregressive conditioner whose layer connections are multiplied element-wise by masks derived from the DAG's adjacency matrix, so each output variable can only see its causal parents. Stacking C-MADEs in series gives MACAW, an invertible affine normalizing flow with a strictly triangular Jacobian; the masks make the flow's density factorization equal to the causal factorization, so $p(x)=\prod_i p(x_i\mid\pi(x_i))$ holds. Kernel PCA first projects 180x180 MRI slices into latent vectors (1500 components, processed in groups of 60) so the flow operates at tractable dimension, and the KPCA preimage reconstructs images from generated latents.

What would settle it

Generate data from a known structural causal model that includes an unobserved confounder or one extra edge, fit MACAW using the graph assumed in the paper, and compute counterfactual queries in the same way as the synthetic experiment (compare predicted counterfactual values to ground-truth values obtained by substituting the intervened value into the true structural equations). If the predicted counterfactual means deviate systematically beyond noise, the causal claims fail.

Watch

Extended reading notes

Core claim

The central claim is that a causal directed acyclic graph can be baked into a normalizing flow by masking each autoregressive conditioner's input connections according to the adjacency matrix. Because the flow's autoregressive factorization $p(x)=\prod_i p(x_i\mid\pi(x_i))$ then coincides with the causal Markov factorization, the single flow is simultaneously a density estimator, an interventional sampler (set a parent to a value and run the backward flow), and a counterfactual engine (abduct the noise with the forward flow, intervene in latent space, predict with the backward flow). The paper demonstrates the first such single-flow encoding of a non-trivial causal structure and uses it to produce 2D counterfactual brain slices in which increasing age enlarges the ventricles and alters sulci, matching established aging anatomy. It also derives exact Bayesian classification from the same density, reporting a mean absolute error of 5.047 years for age prediction on a 60-latent setup.

Load-bearing premise

The entire causal interpretation rests on the pre-specified graph being the true data-generating process (age and sex cause BMI, and all three cause the image latents, with no unobserved confounders), and on the 60-latent subgroups being conditionally independent given those parents; the paper does not test the latter.

Editorial extensions

If this is right

A single trained MACAW model replaces separate conditional generators, classifiers, and counterfactual engines: one forward pass yields the joint density, and one backward pass with fixed parents yields interventional samples.
Counterfactual slices for interventions such as setting age to 75 localize subject-specific structural changes, so the model can serve as an individual-level digital twin for exploring aging effects without collecting longitudinal data.
Bayesian classification comes from the same density estimate, giving posterior distributions over age (or any parent variable) rather than a single point prediction, at the cost of one forward pass per candidate class.
The masking construction is graph-agnostic: adding more causal variables only requires a larger adjacency matrix, so the approach scales in principle beyond the three-variable graph tested here.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because the masking recipe never uses the meaning of age, sex, or BMI, it should transfer to other confounders such as scanner site or disease severity; interventional sampling on scanner identity could then serve as data harmonization, though the paper does not test that.
The split of the 1500 KPCA latents into independently trained groups of 60 is a place where the causal factorization is only enforced within a group; whether causal claims survive across groups is an open question a reader could probe by comparing cross-group counterfactual consistency.
The preimage step from latent to pixel space is a potential fidelity bottleneck; replacing KPCA with an invertible dimensionality reduction inside the flow, which the paper names as future work, would likely be needed before 3D volumes become practical.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Good paper to know about, though the central claim needs qualification.

What's actually new: the C-MADE conditioner, a masked autoencoder that enforces an arbitrary DAG in a single pass, stacked into a normalizing flow. That's a clean and useful extension of MADE to non-autoregressive graphs. The synthetic experiment is the best part: they define a five-variable SEM, train MACAW, and show that counterfactual queries recover the ground truth structural equations almost exactly. That is real evidence the masking does what it claims.

The UK Biobank work is suggestive but shakier. The stress-test note got the key problem right: Section 4.2.2 says training all 1500 KPCA latents in one network didn't converge, so they split the latents into 25 groups of 60 and trained a separate MACAW for each. That is a conditional independence assumption across latent blocks given age, sex, and BMI. KPCA decorrelation is linear and unconditional; it doesn't justify conditional independence of 60-dimensional blocks. The paper never lists this as a limitation. And because counterfactuals are generated independently per block and then reassembled, the resulting image is not the output of a single flow. So the 'single flow' claim in the abstract and Section 1 is only true for the synthetic five-variable case, not for the imaging application that motivates the paper.

Other soft spots, in proportion: the realism baseline is Gaussian blur, which is a low bar; the effectiveness metric is MAE of a surrogate age regressor, reported without error bars; and there's no comparison against existing causal counterfactual models like DSCM or HVAE on the same data. All of these are fixable. The factorization issue is the load-bearing one.

Who this is for: researchers working on causal generative models for medical images, especially those who want a concrete example of masked flows for intervention and counterfactual queries. They should read the synthetic experiment carefully and be cautious about the imaging results.

Recommendation: send it to peer review. The method is well-motivated, the synthetic evidence is solid, and the imaging flaws are addressable with more experiments or a clear statement of the factorization assumption. A good referee can push them to test the independence assumption or use a single flow.

Referee Report

4 major / 5 minor

Summary. The paper proposes MACAW, a causal generative model built from normalizing flows with causally masked autoencoders (C-MADE), and claims that a single flow can encode a pre-specified causal DAG and support association, intervention, and counterfactual queries. The method is evaluated first on a synthetic five-variable structural equation model and then on UK Biobank T1-weighted MRI slices (23,692 participants) using age, sex, and BMI as causal parents of 1500 KPCA latent features. The authors report interventional sampling, counterfactual image generation with FID-based realism and a surrogate-classifier-based effectiveness evaluation, and Bayesian age classification with uncertainty.

Significance. If the claims hold, MACAW is a meaningful step toward integrating causal structure into normalizing flows for image data, offering invertible counterfactual generation and Bayesian classification in a single framework. The synthetic experiment is a genuine strength: it uses known structural equations and shows that the flow can match the SEM distribution and reproduce the expected counterfactual changes, and the code for the toy dataset is provided. However, the significance is limited by the gap between the central 'single flow' claim and the actual UK Biobank implementation, which trains 25 separate flows, and by an untested conditional-independence assumption that is load-bearing for the joint image density and counterfactual coherence. The causal conclusions are also conditional on an assumed DAG, which the authors acknowledge.

major comments (4)

[Section 1 and Section 4.2.2] The central claim that 'a single normalizing flow model coupled with masked autoencoders is sufficient to efficiently model complex causal structures' is not supported by the UK Biobank experiments. Section 4.2.2 states that training all 1500 latents with one MACAW network did not converge, so the latents were divided into subgroups of 60 and a separate MACAW model was trained for each group, with likelihoods optimized individually. The generative sampling and counterfactual inference are then carried out per model and recombined only through the KPCA preimage. The synthetic experiment in Section 4.1 uses one flow with five variables, so it does not demonstrate scaling of a single flow to the image setting. The claim should be qualified, or the image experiment should be redesigned to use a single flow.
[Section 4.2.2 and Section 3.3] Training 25 independent MACAW models on 60-dimensional latent subgroups is equivalent to assuming conditional independence of the subgroups given age, sex, and BMI, i.e., p(latents | age, sex, BMI) = product over g of p(latents_g | age, sex, BMI). This assumption is load-bearing for the joint density and for counterfactual images obtained by independently inverting each flow, but it is not stated or tested. The KPCA orthogonality cited in Section 3.3 provides only unconditional linear decorrelation and does not imply conditional independence given the causal parents. If residual cross-subgroup dependencies exist, for example because global brain size or ventricular volume is captured by several KPCA components, the factorized model is misspecified and the reconstructed counterfactual images do not correspond to a single coherent SCM. The authors should test this assumption, for instance by measuring residual dependencies between subgroups after conditioning on age, sex, and BMI, or by comparing the factorized likelihood against a joint model on a subset of latents.
[Section 4.2.4, Table 2] The effectiveness evaluation uses a single surrogate classifier (SFCN) with no confidence intervals or statistical comparison across the CF conditions, and the realism baseline is only Gaussian blur. The FID values in Table 2 show that the counterfactual images (FID 2.09 to 11.78) are substantially less realistic than the held-out test set (FID 0.49), yet the text interprets them as superior realism relative to blurred images. The claim that the model accurately encodes causal reasoning would be better supported by reporting uncertainty in the MAE and FID estimates, by including a non-causal generative baseline such as a conditional flow or VAE that ignores the DAG, and by applying the axiomatic counterfactual soundness metrics of Monteiro et al. [22], which are cited but not used.
[Section 4.2.5] The Bayesian classification result is computed with the first 60-latent model only, as stated in Section 4.2.5, and therefore does not exercise the full 1500-latent image representation. The reported MAE of 5.047 with standard deviation 0.052 is thus not a classification accuracy of the complete MACAW image model. If the claim is that the model can classify age from images, the classification should be performed using all latent subgroups, or the text should clearly state that this is a proof-of-concept on a subset of latents.

minor comments (5)

[Section 3.2.2, Algorithm 1] The notation in Algorithm 1 is confusing and contains a typo: step 2(a) writes 'zobs_i,xj←α ← T(α, xobs_pi(j))' but the variable index i is not defined in that line, and step 2(b) uses an inconsistent subscript. Please rewrite the algorithm with consistent indexing.
[Figure 2] The caption 'Cf s Ground truth' and the axis label 'x2 Cf alues' appear to have missing characters; the figure should be checked and the caption should be spelled out.
[Section 4.2.1] The paper states that sex information was obtained from the 'genetic sex' data field (22001); the difference between genetic sex and self-reported sex should be clarified, because the causal graph in Figure 3 treats sex as a cause of BMI and of brain images.
[Section 5] The limitations paragraph acknowledges reliance on a pre-defined causal graph but does not mention the conditional independence of the latent subgroups. Given that the factorization is a core assumption of the image pipeline, it should be explicitly listed as a limitation.
[Section 3.3] The phrase 'leading us to assume that this makes the optimization faster' is vague; please state more precisely what is assumed and why.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the synthetic counterfactual check uses external SEM ground truth, the UK Biobank causal graph is an acknowledged input assumption rather than a derived result, and the Bayesian classification is a standard use of the fitted density. The latent-subgroup factorization is a modeling limitation, not a circular reduction.

full rationale

The paper's derivation chain does not reduce to its own inputs. In the synthetic experiment (Section 4.1.4), counterfactual values are compared against ground truth computed directly from the known structural equations (x4 = 6*x2*x3_obs + n4), which is an external reference obtained from the data-generating process, not from the fitted MACAW model; the fact that the model was trained on samples from that SEM does not make the comparison circular, since the counterfactual query still requires the learned flow to correctly perform abduction and prediction. The UK Biobank experiment assumes the causal DAG in Figure 3, and Section 5 explicitly states that 'reliance on a pre-defined causal graph' is a limitation; assuming a graph is an input to the method, but the generated counterfactual images and the claimed aging-related structural changes are not equal to that graph by construction and are checked against external metrics (FID and a separately trained SFCN classifier). The Bayesian classification in Section 4.2.5 uses the standard Bayes rule p(c=ci|f) = p(c=ci,f) / sum_c p(c=cj,f) with likelihoods from the trained flow; this is a normal use of an explicit density estimator, not the renaming of a fitted parameter as a prediction. The claimed factorization of the 1500 KPCA latents into separate 60-dimensional models (Section 4.2.2) is an additional conditional-independence assumption and a possible misspecification, but it is not a step where an output is equivalent to an input by definition. The self-citations [13,14] are used only to describe prior related work in the same group and do not carry the load of any derivation or uniqueness claim. Overall, the central results are supported by external benchmarks and stated assumptions, so no circularity is present.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The method itself is a standard maximum-likelihood generative model; the main assumptions are the pre-specified causal graph, the conditional independence of latent subgroups, the adequacy of the 2D slice, and the standard Markov/faithfulness conditions. There are no invented physical entities.

free parameters (5)

KPCA total components and subgroup size = 1500 total; 60 per MACAW model
Section 4.2.2: training 1500 latents in one model did not converge; subgroup size chosen by hand to get better likelihood.
Prior probabilities for sex and age = Bernoulli and categorical estimated from training data
Section 4.2.2: priors for sex and age are determined from training distribution, making the class prior fitted to the data.
KPCA polynomial kernel degree = 3
Section 3.3 selects degree-3 polynomial kernel; no sensitivity analysis reported.
Network architecture sizes = Synthetic: 10 C-MADEs, 3 hidden layers, 15 neurons
Section 4.1.2; no search or ablation reported.
Gaussian blur sigma for realism baseline = 1
Section 4.2.4: blurred baseline uses standard deviation 1, an arbitrary comparator.

assumptions (5)

standard math Markov condition and faithfulness hold for the assumed DAG.
Section 2.1 invokes them to justify factorizing the joint density according to the causal graph.
domain assumption The UK Biobank causal graph (Figure 3) is correct and there are no unobserved confounders.
Section 4.2.2 and Figure 3; acknowledged in Section 5 as reliance on a pre-defined causal graph.
ad hoc to paper The 60-latent subgroups are conditionally independent given age, sex, and BMI.
Section 4.2.2 trains separate MACAW models per subgroup and combines them; this factorization is never tested.
domain assumption The selected 2D axial slice captures the causal effects of age, sex, and BMI on brain structure.
Section 4.2.1 chooses the lateral ventricular slice for all analyses.
standard math Normalizing flow change-of-variables formula and invertibility conditions hold.
Section 2.2; standard background, not original.

how reviews work

0 comments

Cite this review

Pith. "Pith review of MACAW: A Causal Generative Model for Medical Imaging." pith.science (2026). https://pith.science/paper/M7NQMSYG

@misc{pith2026241202900,
  author       = {Pith},
  title        = {Pith review of: MACAW: A Causal Generative Model for Medical Imaging},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/M7NQMSYG}},
  note         = {Machine review of arXiv:2412.02900}
}

read the original abstract

Although deep learning techniques show promising results for many neuroimaging tasks in research settings, they have not yet found widespread use in clinical scenarios. One of the reasons for this problem is that many machine learning models only identify correlations between the input images and the outputs of interest, which can lead to many practical problems, such as encoding of uninformative biases and reduced explainability. Thus, recent research is exploring if integrating a priori causal knowledge into deep learning models is a potential avenue to identify these problems. This work introduces a new causal generative architecture named Masked Causal Flow (MACAW) for neuroimaging applications. Within this context, three main contributions are described. First, a novel approach that integrates complex causal structures into normalizing flows is proposed. Second, counterfactual prediction is performed to identify the changes in effect variables associated with a cause variable. Finally, an explicit Bayesian inference for classification is derived and implemented, providing an inherent uncertainty estimation. The feasibility of the proposed method was first evaluated using synthetic data and then using MRI brain data from more than 23000 participants of the UK biobank study. The evaluation results show that the proposed method can (1) accurately encode causal reasoning and generate counterfactuals highlighting the structural changes in the brain known to be associated with aging, (2) accurately predict a subject's age from a single 2D MRI slice, and (3) generate new samples assuming other values for subject-specific indicators such as age, sex, and body mass index. The code for a toy dataset is available at the following link: https://github.com/vibujithan/macaw-2D.git.

Figures

Figures reproduced from arXiv: 2412.02900 by the authors.

**Figure 2.** (a) The sum of the absolute difference between each varia [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗

**Figure 3.** (a) Predefined causal graph for the UKBB dataset, inco [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: Results from two types of generative sampling: unconditio [PITH_FULL_IMAGE:figures/full_fig_p017_4.png]

**Figure 5.** Figure 5: Illustration of the outcomes of counterfactual queries [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]

**Figure 6.** Figure 6: Results from counterfactual queries simulating brain agin [PITH_FULL_IMAGE:figures/full_fig_p019_6.png]

**Figure 7.** Figure 7: The figure displays how the effectiveness (MAE) of the gen [PITH_FULL_IMAGE:figures/full_fig_p020_7.png]

**Figure 8.** Figure 8: Posterior estimated by the model for each age, with the a [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Causal Transfer in Medical Image Analysis
cs.CV 2026-03 accept novelty 5.0 of 10

Causal Transfer Learning unifies structural causal models, invariant risk minimisation and counterfactuals with transfer learning to produce domain-robust medical image models.

Reference graph

Works this paper leans on

32 extracted references · 14 canonical work pages · cited by 1 Pith paper

[22]

Monteiro, F

M. Monteiro, F. D. S. Ribeiro, N. Pawlowski, D. C. Castro, B. Gloc ker, Measuring axiomatic soundness of counterfactual image models (2 023). arXiv:2303.01274

arXiv
[1]

S. J. MacEachern, N. D. Forkert, Machine learning for precision medicine, Genome 64 (4) (2021) 416–425. doi:10.1139/gen-2020-0131

work page doi:10.1139/gen-2020-0131 2021
[2]

A. J. Winder, E. A. Stanley, J. Fiehler, N. D. Forkert, Challenges and potential of artiﬁcial intelligence in neuroradiology, Clinical Neurora di- ology (2024) 1–13

work page 2024
[3]

D. C. Castro, I. Walker, B. Glocker, Causality matters in medical imag- ing, Nature Communications 11 (1) (2020) 3673, number: 1 Publishe r: Nature Publishing Group. doi:10.1038/s41467-020-17478-w

doi:10.1038/s41467-020-17478-w 2020
[4]

L. L. Vercio, K. Amador, J. J. Bannister, S. Crites, A. Gutierre z, M. E. MacDonald, J. Moore, P. Mouches, D. Rajashekar, S. Schimert, N . Sub- banna, A. Tuladhar, N. Wang, M. Wilms, A. Winder, N. D. Forkert, Su - pervised machine learning tools: a tutorial for clinicians, Journal of Neu- ral Engineering 17 (6) (2020) 062001. doi:10.1088/1741-2552/abbff2. 23

work page doi:10.1088/1741-2552/abbff2 2020
[5]

Souza, A

R. Souza, A. Winder, E. A. M. Stanley, V. Vigneshwaran, M. Cama - cho, R. Camicioli, O. Monchi, M. Wilms, N. D. Forkert, Identifying biases in a multicenter mri database for parkinson’s disease classiﬁ- cation: Is the disease classiﬁer a secret site classiﬁer?, IEEE Jour - nal of Biomedical and Health Informatics 28 (4) (2024) 2047–2054 . doi:10.1109/JBHI...

arXiv 2024
[6]

Sanchez, J

P. Sanchez, J. P. Voisey, T. Xia, H. I. Watson, A. Q. O’Neil, S. A. T saf- taris, Causal machine learning for healthcare and precision medicine , Royal Society Open Science 9 (8) (2022) 220638, publisher: Royal Soci- ety. doi:10.1098/rsos.220638

doi:10.1098/rsos.220638 2022
[7]

Pearl, The Causal Foundations of Structural Equation Mode ling:, Tech

J. Pearl, The Causal Foundations of Structural Equation Mode ling:, Tech. rep., Defense Technical Information Center, Fort Belvoir, V A (Feb. 2012). doi:10.21236/ADA557445

work page doi:10.21236/ada557445 2012

Show all 32 references

[8]

Pawlowski, D

N. Pawlowski, D. Coelho de Castro, B. Glocker, Deep Structural Causal Models for Tractable Counterfactual Inference, in: Advances in Neural Information Processing Systems, Vol. 33, Curran Associates, In c., 2020, pp. 857–869

2020
[9]

Xia, K.-Z

K. Xia, K.-Z. Lee, Y. Bengio, E. Bareinboim, The causal- neural connection: Expressiveness, learnability, and inference ( 2022). arXiv:2107.00793

2022 arXiv
[10]

F. D. S. Ribeiro, T. Xia, M. Monteiro, N. Pawlowski, B. Glocker, Hig h Fidelity Image Counterfactuals with Probabilistic Causal Models, in: Proceedings of the 40th International Conference on Machine Le arning, PMLR, 2023, pp. 7390–7425, iSSN: 2640-3498

2023
[11]

W. Peng, T. Xia, F. D. S. Ribeiro, T. Bosschieter, E. Adeli, Q. Zha o, B. Glocker, K. M. Pohl, Latent 3d brain mri counterfactual (2024 ). arXiv:2409.05585

2024
[12]

Sanchez, S

P. Sanchez, S. A. Tsaftaris, Diﬀusion causal models for count erfactual estimation (2022). arXiv:2202.10166

2022 arXiv
[13]

Wilms, J

M. Wilms, J. J. Bannister, P. Mouches, M. E. MacDonald, D. Ra- jashekar, S. Langner, N. D. Forkert, Invertible Modeling of Bidire ctional Relationships in Neuroimaging With Normalizing Flows: Application 24 to Brain Aging, IEEE Transactions on Medical Imaging 41 (9) (2022) 233...

2022
[14]

J. J. Bannister, M. Wilms, J. D. Aponte, D. C. Katz, O. D. Klein, F. P. J. Bernier, R. A. Spritz, B. Hallgrimsson, N. D. Forkert, A Dee p Invertible 3-D Facial Shape Model for Interpretable Genetic Synd rome Diagnosis, IEEE journal of biomedical and health informatics 26 (7) (...

2022
[15]

Peters, D

J. Peters, D. Janzing, B. Sch¨ olkopf, Elements of causal infe rence: foun- dations and learning algorithms, Adaptive computation and machine learning series, The MIT Press, Cambridge, Massachuestts, 2017

2017
[16]

Glymour, K

C. Glymour, K. Zhang, P. Spirtes, Review of Causal Discovery M ethods Based on Graphical Models, Frontiers in Genetics 10 (2019)

2019
[17]

Kobyzev, S

I. Kobyzev, S. J. Prince, M. A. Brubaker, Normalizing Flows: An Intro- duction and Review of Current Methods, IEEE Transactions on Pat tern Analysis and Machine Intelligence 43 (11) (2021) 3964–3979, confe rence Name: IEEE Transactions on Pattern Analysis and Machine Intelli- ...

2021
[18]

Wehenkel, G

A. Wehenkel, G. Louppe, Graphical Normalizing Flows, in: Procee d- ings of The 24th International Conference on Artiﬁcial Intelligenc e and Statistics, PMLR, 2021, pp. 37–45, iSSN: 2640-3498

2021
[19]

Khemakhem, R

I. Khemakhem, R. Monti, R. Leech, A. Hyvarinen, Causal Auto regres- sive Flows, in: Proceedings of The 24th International Conference on Artiﬁcial Intelligence and Statistics, PMLR, 2021, pp. 3520–3528, iSSN: 2640-3498

2021
[20]

Germain, K

M. Germain, K. Gregor, I. Murray, H. Larochelle, MADE: Maske d Au- toencoder for Distribution Estimation, in: Proceedings of the 32nd In- ternational Conference on Machine Learning, PMLR, 2015, pp. 88 1–889, iSSN: 1938-7228

2015
[21]

Papamakarios, T

G. Papamakarios, T. Pavlakou, I. Murray, Masked Autoregre ssive Flow for Density Estimation, in: Advances in Neural Information Proces sing Systems, Vol. 30, Curran Associates, Inc., 2017. 25

2017
[23]

Melistas, N

T. Melistas, N. Spyrou, N. Gkouti, P. Sanchez, A. Vlontzos, G. Papanas- tasiou, S. A. Tsaftaris, Benchmarking Counterfactual Image Ge neration, arXiv:2403.20287 [cs] (Mar. 2024). doi:10.48550/arXiv.2403.20287

2024 doi arXiv
[24]

Szegedy, V

C. Szegedy, V. Vanhoucke, S. Ioﬀe, J. Shlens, Z. Wojna, Ret hinking the Inception Architecture for Computer Vision, arXiv:1512.00567 [cs] version: 3 (Dec. 2015). doi:10.48550/arXiv.1512.00567

2015 doi arXiv
[25]

K. L. Miller, F. Alfaro-Almagro, N. K. Bangerter, D. L. Thomas, E. Yacoub, J. Xu, A. J. Bartsch, S. Jbabdi, S. N. Sotiropoulos, J. L. R. Andersson, L. Griﬀanti, G. Douaud, T. W. Okell, P. Weale, I. Dragonu, S. Garratt, S. Hudson, R. Collins, M. Jenkinson, P. M. Matthews, S. M....

2016 doi
[26]

Rohlﬁng, N

T. Rohlﬁng, N. M. Zahr, E. V. Sullivan, A. Pfeﬀerbaum, The SRI2 4 multichannel atlas of normal adult human brain structure, Human B rain Mapping 31 (5) (2010) 798–819. doi:10.1002/hbm.20906

2010 doi
[27]

B. B. Avants, N. J. Tustison, G. Song, P. A. Cook, A. Klein, J. C . Gee, A reproducible evaluation of ANTs similarity metric performance in brain image registration, NeuroImage 54 (3) (2011) 2033–2044. doi:10.1016/j.neuroimage.2010.09.025

2011 doi
[28]

Lotze, M

M. Lotze, M. Domin, F. H. Gerlach, C. Gaser, E. Lueders, C. O. Schmidt, N. Neumann, Novel ﬁndings from 2,838 Adult Brains on Sex Diﬀerences in Gray Matter Brain Volume, Scientiﬁc Reports 9 (1) (2019) 1671, number: 1 Publisher: Nature Publishing Group. doi:10.1038/s41598-018-38239-2

2019 doi
[29]

D. Beck, A. G. de Lange, M. L. Pedersen, D. Alnæs, I. I. Maximo v, I. Voldsbekk, G. Richard, A. Sanders, K. M. Ulrichsen, E. S. Dørum , K. K. Kolsk ˚ ar, E. A. Høgestøl, N. E. Steen, S. Djurovic, O. A. An- dreassen, J. E. Nordvik, T. Kaufmann, L. T. Westlye, Cardiometa bolic 2...

2021 doi
[30]

H. Peng, W. Gong, C. F. Beckmann, A. Vedaldi, S. M. Smith, Accu rate brain age prediction with lightweight deep neural networks, Medical Image Analysis 68 (2021) 101871. doi:10.1016/j.media.2020.101871

2021
[31]

M. J. Kusner, J. R. Loftus, C. Russell, R. Silva, Counterfactu al fairness (2018). arXiv:1703.06856

2018 arXiv
[32]

Nemirovsky, N

D. Nemirovsky, N. Thiebaut, Y. Xu, A. Gupta, CounteRGAN: Ge n- erating Realistic Counterfactuals with Residual Generative Advers arial Nets, arXiv:2009.05199 [cs, stat] (May 2021). 27

2009 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [22]

Monteiro, F

M. Monteiro, F. D. S. Ribeiro, N. Pawlowski, D. C. Castro, B. Gloc ker, Measuring axiomatic soundness of counterfactual image models (2 023). arXiv:2303.01274

arXiv

[2] [1]

S. J. MacEachern, N. D. Forkert, Machine learning for precision medicine, Genome 64 (4) (2021) 416–425. doi:10.1139/gen-2020-0131

work page doi:10.1139/gen-2020-0131 2021

[3] [2]

A. J. Winder, E. A. Stanley, J. Fiehler, N. D. Forkert, Challenges and potential of artiﬁcial intelligence in neuroradiology, Clinical Neurora di- ology (2024) 1–13

work page 2024

[4] [3]

D. C. Castro, I. Walker, B. Glocker, Causality matters in medical imag- ing, Nature Communications 11 (1) (2020) 3673, number: 1 Publishe r: Nature Publishing Group. doi:10.1038/s41467-020-17478-w

doi:10.1038/s41467-020-17478-w 2020

[5] [4]

L. L. Vercio, K. Amador, J. J. Bannister, S. Crites, A. Gutierre z, M. E. MacDonald, J. Moore, P. Mouches, D. Rajashekar, S. Schimert, N . Sub- banna, A. Tuladhar, N. Wang, M. Wilms, A. Winder, N. D. Forkert, Su - pervised machine learning tools: a tutorial for clinicians, Journal of Neu- ral Engineering 17 (6) (2020) 062001. doi:10.1088/1741-2552/abbff2. 23

work page doi:10.1088/1741-2552/abbff2 2020

[6] [5]

Souza, A

R. Souza, A. Winder, E. A. M. Stanley, V. Vigneshwaran, M. Cama - cho, R. Camicioli, O. Monchi, M. Wilms, N. D. Forkert, Identifying biases in a multicenter mri database for parkinson’s disease classiﬁ- cation: Is the disease classiﬁer a secret site classiﬁer?, IEEE Jour - nal of Biomedical and Health Informatics 28 (4) (2024) 2047–2054 . doi:10.1109/JBHI...

arXiv 2024

[7] [6]

Sanchez, J

P. Sanchez, J. P. Voisey, T. Xia, H. I. Watson, A. Q. O’Neil, S. A. T saf- taris, Causal machine learning for healthcare and precision medicine , Royal Society Open Science 9 (8) (2022) 220638, publisher: Royal Soci- ety. doi:10.1098/rsos.220638

doi:10.1098/rsos.220638 2022

[8] [7]

Pearl, The Causal Foundations of Structural Equation Mode ling:, Tech

J. Pearl, The Causal Foundations of Structural Equation Mode ling:, Tech. rep., Defense Technical Information Center, Fort Belvoir, V A (Feb. 2012). doi:10.21236/ADA557445

work page doi:10.21236/ada557445 2012