Pith. sign in

REVIEW 3 major objections 6 minor 54 references

scBIT: Integrating Single-cell Transcriptomic Data into fMRI-based Prediction for Alzheimer's Disease Diagnosis

T0 review · 3 major / 6 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read Adding single-cell RNA data to an fMRI-based model improves Alzheimer's diagnosis accuracy by 3.39 percentage points in binary classification and 26.59 points in five-stage severity classification.

desk verdict A genuinely novel snRNA+fMRI architecture whose reported accuracy gains are likely inflated by test-set leakage in the pre-trained encoders; worth reviewing, but not credible as presented. read the letter →

arxiv 2502.02630 v1 pith:F637HR7C submitted 2025-02-04 q-bio.QM cs.AIcs.LG

Yu-An Huang , Yao Hu , Yue-Chao Li , Xiyue Cao , Xinyuan Li , Kay Chen Tan , Zhu-Hong You , Zhi-An Huang This is my paper · ORCID
classification q-bio.QMcs.AIcs.LG
keywords single-cellimagingtranscriptomicsAlzheimer'sdiseasediagnosiscross-modalcontrastivelearninggraphneuralnetworkfunctionalMRIbrainregion-geneassociationsnRNAmixture-of-experts
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 introduces scBIT, a computational pipeline that uses single-nucleus RNA (snRNA) data to improve Alzheimer's disease prediction from resting-state fMRI alone. Its central claim is that adding this second modality, even though the snRNA and fMRI come from different people, raises classification accuracy by 3.39 percentage points in a binary AD-versus-control task and by 26.59 percentage points in a five-stage severity task. This matters because fMRI is widely collected while single-cell transcriptomes carry molecular detail that fMRI cannot see; if the pairing is genuine, existing fMRI-only cohorts could be re-analyzed for more accurate staging and for molecularly informed biomarker discovery. scBIT avoids the need for paired samples by linking the two datasets through age, sex, disease-state, and genetic similarity, and it makes its reasoning traceable by mapping brain regions to cell-type-specific gene subgraphs.

What carries the argument

The load-bearing machinery is a 'cell bagging' scheme that pools nuclei of the same cell type into gene networks, a self-explainable graph neural network that distills those networks into cell-type prototype subgraphs, and four pre-trained fMRI encoders whose cosine similarities to the subgraph embeddings serve as cross-modal attention scores. A mixture-of-experts layer then learns per-patient weights for the four attention scores and concatenates the fused score with the fMRI embedding for classification. The whole design exists to give the fMRI classifier access to single-cell molecular information without requiring paired multi-modal data.

What would settle it

Re-run scBIT with the test fold's fMRI data completely excluded from every pre-training stage; if the reported gains of 3.39% and 26.59% collapse, the improvement is an artifact of test-set statistics rather than the snRNA modality.

Watch

Extended reading notes

Core claim

The paper's core discovery is that cross-modal attention can transfer information from single-cell gene-expression data into an fMRI classifier without requiring the same individual to have both measurements. Each snRNA sample is converted into 90 prototype gene subgraphs spanning 15 cell types, four fMRI encoders are trained against age, sex, clinical-state, and genetic similarity labels, and the resulting attention scores are fused with brain-region embeddings by a mixture-of-experts layer before diagnosis. In a ten-fold cross-validation study on a public Alzheimer's cohort, the full model reaches 0.958 accuracy for binary AD classification and sharply improves five-class severity classification, with the authors reporting that genetic similarity is the strongest single matching signal and that accuracy rises as more similarity types are included. The authors present this as the first computational model to integrate single-cell RNA data with MRI data for Alzheimer's prediction.

Load-bearing premise

The evaluation assumes that masking the disease-state similarity label for the test set is enough to prevent leakage, even though the test fold's fMRI features are included in pre-training.

Editorial extensions

If this is right

  • Alzheimer's prediction from fMRI alone can be improved using an external single-cell transcriptomic atlas, even when no individual appears in both modalities.
  • The largest gains appear in five-stage severity classification, suggesting the snRNA signal is especially informative for disease staging rather than mere detection.
  • Genetic similarity is the dominant matching signal, so genotype-aware sample pairing may matter more than demographic matching for cross-modal transfer.
  • The learned attention maps give testable hypotheses about which brain regions and cell-type-specific gene subgraphs are implicated in AD.
  • Existing public fMRI datasets could in principle be re-analyzed with this method to obtain molecularly informed diagnostic scores without new data collection.

Reading between the lines

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

  • Because the test fold's fMRI features participate in pre-training, the reported accuracy gain could partly reflect the encoder learning test-set statistics; a fully held-out fMRI split would be needed to confirm the gain is caused by the snRNA information.
  • If the leakage concern is resolved, the same unpaired cross-modal matching scheme could be applied to other conditions that have both single-cell atlases and neuroimaging cohorts.
  • The specific brain-region–gene-subgraph pairs highlighted by the attention weights could be validated as biomarkers by measuring the same genes or regions in independent AD cohorts.
  • The large five-class improvement suggests that the auxiliary snRNA signal may help refine disease staging even when binary detection is already saturated.
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.

Referee Report

3 major / 6 minor

Summary. The paper proposes scBIT, a three-stage framework that integrates single-nucleus RNA-seq (snRNA) with resting-state fMRI for Alzheimer's disease (AD) diagnosis. Stage 1 converts snRNA data into cell-type-specific gene subgraph embeddings using a self-explainable graph neural network; Stage 2 trains four fMRI encoders on age, sex, clinical-state, and genetic similarities to compute cross-modal attention scores against the snRNA prototypes; Stage 3 fuses those attention scores with fMRI embeddings for classification. The authors report ten-fold cross-validation on ADNI, claiming that incorporating snRNA improves binary classification accuracy by 3.39% and five-class classification accuracy by 26.59%, and they release code on GitHub and Zenodo.

Significance. If the empirical results were sound, scBIT would be a genuinely novel cross-modal diagnostic pathway that brings single-cell transcriptomics into fMRI-based AD prediction, with interpretable brain-region–gene associations. The paper has clear strengths: it uses public datasets, releases code and detailed instructions, and builds the interpretability machinery (ProtGNN-style subgraph extraction and cell-type-level attention) directly into the model. However, the central empirical claim rests on an evaluation protocol that appears to include test-set fMRI features during pre-training, which undermines the reported accuracy gains. The contribution is valuable in conception, but the current evidence does not establish the claimed snRNA benefit.

major comments (3)
  1. [Section 4.2 and Stage 2 of Section 3.3] The statement that 'to avoid the data leakage problem, we mask the disease state similarity corresponding to the testing set during the training of the pre-trained encoder' is not sufficient to prevent leakage. The text does not say that the test fold's fMRI inputs are excluded from pre-training the four encoders E_rm^{pt,s} in Eq. (12); it only masks one label type. Consequently, the age, sex, and gene encoders can be trained with test fMRI features and their labels, and even the state encoder can see test fMRI inputs. Because the attention scores in Eq. (14) are computed with these encoders and then fed into the classifier in Eq. (16), test-set statistics can enter the predictions. The load-bearing claim of snRNA-driven improvement therefore is not established; the evaluation must be redone with the test fold excluded from all pre-training.
  2. [Section 4.2, Fig. 3(a)] The 'without snRNA' baseline is not a controlled ablation of the snRNA modality. It omits the entire Stage 2 pretrained-encoder and attention pathway, so the reported 3.39% and 26.59% gains cannot be attributed specifically to the snRNA-derived prototypes; they may instead come from the additional pretrained features, particularly the state-similarity encoder. The paper should compare scBIT against a baseline that includes the same Stage 2 encoders and attention features but with the snRNA prototypes P replaced by a non-informative or randomly permuted prototype set, so that the snRNA contribution is isolated.
  3. [Section 3.3, Eqs. (13) and (16), and Section 4.3] The state-similarity encoder is trained directly on the clinical-state labels, and its attention score a_i^{state} is fed into the classifier predicting those same clinical states. Even under a clean data split, this creates a feature that is essentially a learned encoding of the diagnostic label, and the MoE weight for the state similarity is reported as 11.11% (Fig. 3(h)). The paper should include an ablation that removes the state-similarity attention or replaces it with a non-diagnostic similarity, and should also validate the linear mapping from SEA-AD 'Low/Intermediate/High' to ADNI 'EMCI/MCI/LMCI' categories; without such an ablation, the snRNA modality itself may not be responsible for the observed improvement.
minor comments (6)
  1. [Section 4.1] The subsection title contains a typo: 'Data and materlals' should read 'Data and materials'.
  2. [Eq. (3)] Equation (3) writes d_q = [sim(p_i, h_q) | k = 1 … 90], but the index in the expression should be consistent with the loop variable; presumably it should be i = 1 … 90 or p_k.
  3. [Eq. (13)] The symbol ∤(·) is used for embedding normalization but is never defined; please use a standard notation such as ‖·‖-normalized or define the operation explicitly.
  4. [Section 4.2] The paper does not report the number of ADNI fMRI subjects or the class counts for the binary and five-class tasks; these numbers are essential for interpreting the reported accuracies and standard deviations.
  5. [Table 1] The comparison with prior methods in Table 1 relies on best results reported in the literature, which may use different data splits, patient selections, and preprocessing; the text should clarify whether metrics are matched on the same subjects and folds.
  6. [Fig. 3(b) and Fig. 3(c)] The subgroup accuracies for gender and age are reported without standard deviations or sample sizes; please include these to support the subgroup comparison.

Circularity Check

2 steps flagged · score 6.0 of 10

The reported snRNA-driven accuracy gain is confounded by test-set pretraining of the cross-modal encoders; the state-similarity attention also injects target-label-derived signal into the final classifier.

  1. fitted input called prediction [Section 3.3 (Stage 2, Eq. 13-14; Stage 3, Eq. 16) and Section 4.2]
    "Then, supervised contrastive learning, ℒ𝑐𝑙^𝑠, is leveraged to train 𝐸𝑟𝑚^{𝑝𝑡,𝑠}... a𝑖^𝑠 = 𝑐𝑠𝑖𝑚(𝑃, 𝐸𝑟𝑚^{𝑝𝑡,𝑠}(𝑋𝑖))... ŷ𝑖 = Sigmoid(𝑓(⟨𝑋𝑖′, 𝐴𝑖⟩))... To avoid the data leakage problem, we mask the disease state similarity corresponding to the testing set during the training of the pre-trained encoder."

    The disease-state attention pathway is built by training an fMRI encoder with supervised contrastive learning on clinical-state similarity labels (Eq. 13). Its output is then converted into an attention score (Eq. 14), combined into A_i (Eq. 15), and fed into the classifier that predicts the same clinical state (Eq. 16). Thus the model's diagnostic prediction includes a feature whose construction is directly supervised by the target variable. The masking sentence only hides the test-set state similarity; it does not state that test fMRI inputs are excluded from pretraining, so the reported 'with snRNA' gain is partly a fitted artifact rather than a clean measure of snRNA contribution.

  2. other [Section 3.3 (Stage 2, Eq. 12 and Eq. 15) and Section 4.2 (cross-validation protocol)]
    "For the attention scores for continuous similarity values, scBIT take the training of 𝐸𝑟𝑚^{𝑝𝑡,𝑠} where 𝑠 ∈ {age, gene} as regression tasks by directly using the cross-modal individual similarities as labels. ... To avoid the data leakage problem, we mask the disease state similarity corresponding to the testing set during the training of the pre-trained encoder."

    The pretraining protocol fits the age, sex, and gene ROI encoders on the fMRI dataset using age, sex, and SNP-derived genetic labels, and only the disease-state similarity is masked for the test fold. At inference, these encoders produce attention scores that enter A_i via Eq. 15. Consequently, test-fold age, sex, and APOE-related genetic information is used to train feature extractors whose outputs are later scored on those same test folds. Since these covariates are strong AD predictors, the transductive fit, rather than snRNA content, can explain part or all of the reported accuracy improvement over the Stage-3-only baseline.

full rationale

The paper's headline claim is that adding snRNA data improves fMRI-based AD diagnosis by 3.39% (binary) and 26.59% (five-class). This comparison is between the full scBIT pipeline and a baseline that uses only the Stage-3 ROI encoder. The full pipeline adds four pretrained cross-modal attention scores, and these scores are produced by encoders that, according to the stated protocol, are pretrained on the fMRI dataset while only the disease-state similarity for the test set is masked. The state-similarity encoder in particular is trained with the target clinical labels and its output is then used as an input feature for predicting those same labels. This is not a clean independent test of snRNA value: the accuracy gain can be partially attributed to label-supervised feature construction and to test-set information entering through the age, sex, and gene encoders. No load-bearing self-citation chain was found; the methodological issue is the evaluation protocol itself. The derivation of the GNN subgraph representation and the MoE fusion are internally coherent, but the central numerical claims are not established by the reported experiments.

Assumptions & free parameters 10 free parameters · 5 assumptions · 0 invented entities

The method introduces no new physical entities. It relies on several hand-chosen parameters and domain assumptions: the bagging and threshold choices for gene network construction, the mapping of clinical states to numbers, and the assumption that the four similarity metrics are sufficient to match unpaired individuals. The most consequential assumption is the comparability of clinical state categories across SEA-AD and ADNI, which directly feeds the circularity concern.

free parameters (10)
  • Bag size (nuclei per bag) = 20
    Number of nuclei sampled per bag for gene network construction in Stage 1.
  • PCC edge retention threshold = top 20% of gene-gene pairs
    Edges with highest Pearson correlation coefficient retained to form gene networks.
  • Number of prototypes per cell type = 6
    Total 90 prototypes across 15 cell types in the self-explainable GNN.
  • Lambda_1 = 0.1
    Weight for prototype closeness loss in Eq. 7.
  • Lambda_2 = 0.05
    Weight for prototype separation loss in Eq. 7.
  • Tau (contrastive temperature) = 0.1
    Temperature parameter in supervised contrastive loss (Eq. 13).
  • Epsilon = 1e-4
    Smoothing constant in the similarity computation (Eq. 10).
  • Clinical state mapping = snRNA: {0,0.25,0.5,0.75}, fMRI: {0,0.25,0.5,0.75,1}
    Ad hoc linear mapping of disease severity categories to numeric values for similarity calculation.
  • Sex mapping = male=1, female=-1
    Binary encoding used to compute sex similarity.
  • Age mapping = 90+ treated as 90
    Censored age values are replaced with 90 for similarity computation.
assumptions (5)
  • domain assumption Cell type annotations in the Seattle Alzheimer's Disease Brain Cell Atlas are accurate and consistent across individuals.
    The GNN in Stage 1 is trained to predict these cell type labels; errors would propagate to the learned prototypes.
  • domain assumption Pearson correlation coefficient between gene pairs across 20 nuclei captures biologically meaningful gene interactions.
    The gene network construction relies on PCC-based edge selection to define the graphs.
  • domain assumption Age, sex, clinical state, and genetic similarity are appropriate and sufficient for matching unpaired snRNA and fMRI individuals.
    The cross-modal matching uses only these four similarity metrics to bridge distinct cohorts.
  • ad hoc to paper The clinical state categories in SEA-AD (Not AD, Low, Intermediate, High) are mappable to ADNI categories (CN, EMCI, MCI, LMCI, AD) via a linear numeric scale.
    The mapping 0,0.25,0.5,0.75,1 is introduced without justification and underlies the state similarity.
  • domain assumption The learned prototypes from snRNA data are representative of all snRNA datasets and can be used as a fixed dictionary for fMRI matching.
    The same prototype set P is used for all fMRI samples in attention computation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of scBIT: Integrating Single-cell Transcriptomic Data into fMRI-based Prediction for Alzheimer's Disease Diagnosis." pith.science (2026). https://pith.science/paper/F637HR7C

@misc{pith2026250202630,
  author       = {Pith},
  title        = {Pith review of: scBIT: Integrating Single-cell Transcriptomic Data into fMRI-based Prediction for Alzheimer's Disease Diagnosis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/F637HR7C}},
  note         = {Machine review of arXiv:2502.02630}
}
read the original abstract

Functional MRI (fMRI) and single-cell transcriptomics are pivotal in Alzheimer's disease (AD) research, each providing unique insights into neural function and molecular mechanisms. However, integrating these complementary modalities remains largely unexplored. Here, we introduce scBIT, a novel method for enhancing AD prediction by combining fMRI with single-nucleus RNA (snRNA). scBIT leverages snRNA as an auxiliary modality, significantly improving fMRI-based prediction models and providing comprehensive interpretability. It employs a sampling strategy to segment snRNA data into cell-type-specific gene networks and utilizes a self-explainable graph neural network to extract critical subgraphs. Additionally, we use demographic and genetic similarities to pair snRNA and fMRI data across individuals, enabling robust cross-modal learning. Extensive experiments validate scBIT's effectiveness in revealing intricate brain region-gene associations and enhancing diagnostic prediction accuracy. By advancing brain imaging transcriptomics to the single-cell level, scBIT sheds new light on biomarker discovery in AD research. Experimental results show that incorporating snRNA data into the scBIT model significantly boosts accuracy, improving binary classification by 3.39% and five-class classification by 26.59%. The codes were implemented in Python and have been released on GitHub (https://github.com/77YQ77/scBIT) and Zenodo (https://zenodo.org/records/11599030) with detailed instructions.

Figures

Figures reproduced from arXiv: 2502.02630 by the authors.

Figure 1
Figure 1. Overview of the scBIT framework. 2. Related works Since there are currently no studies combining single-cell data with MRI data for predicting AD, this section focuses on methods that exclusively utilize MRI data for AD classification or prediction. Recent research in AD prediction using neuroimaging data incorporates advanced methodologies that include network techniques and machine learning algorithms. Mao et al. … view at source ↗
Figure 3
Figure 3. Predictive Performance of the scBIT Model and Statistics of the snRNA-fMRI Cross-Modal Dataset. (a) Comparison of prediction performance with and without pre-trained embeddings in (left) binary disease diagnosis tasks and (right) classifying five severity levels of AD. (b) Disease diagnosis predictions under gender division. (c) Disease diagnosis predictions across different age groups. (d) Performance of scBIT with… view at source ↗
Figure 4
Figure 4. Predictive performance of scBIT across different public gene pathway datasets. (a) Venn diagram of gene counts across six gene pathway datasets used. (b) Performance comparison of scBIT using different gene pathway databases. (c) Distribution of gene expression across different cell types in patient and healthy groups using the minimal gene pathway dataset for snRNA data. (d) Distribution of gene subgraph phenotypes… view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: scBIT provides interpretability regarding the correlations among cell type, brain region, and gene networks in AD. (a) Visualization of attention between cell types and brain regions in the patient group learned from scBIT, showing the top 20% of (left) brain regions w…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 52 canonical work pages

  1. [1]

    Genetic sequencing provides detailed information about an individual's genome at a microscopic level, including specific genetic variations and genetic risk -related data [2, 3]

    Introduction Alzheimer’s disease, a progressive neurological disorder impacting cognitive function and daily life, affects an estimated 50 million people worldwide and is diagnosed using genetic sequencing and medical imaging techniques [1]. Genetic sequencing provides detailed information about an individual's genome at a microscopic level, including spe...

    work page

  2. [2]

    Recent research in AD prediction using neuroimaging data incorporates advanced methodologies that include network techniques and machine learning algorithms

    Related works Since there are currently no studies combining single -cell data with M RI data for predicting AD, this section focuses on methods that exclusively utilize MRI data for AD classification or prediction. Recent research in AD prediction using neuroimaging data incorporates advanced methodologies that include network techniques and machine lear...

    work page

  3. [3]

    1" to males and

    Methodology 3.1 Problem formulation In this article, the prediction task of AD diagnosis using snRNA and fMRI multimodal data is designed with snRNA as the helper modality. Through contrastive learning, correlations across different samples from the datasets are constructed, and these correlations are then integrated with fMRI data, which serves as the pr...

    work page

  4. [4]

    In our study, we utilized multiple datasets to provide a comprehensive analysis of AD

    Results 4.1 Data and materlals All datasets used in this work are publicly available. In our study, we utilized multiple datasets to provide a comprehensive analysis of AD. Functional MRI data were sourced from the ADNI, which includes imaging data along with demographic information such as age, sex, and disease status of the participants. This data can b...

    work page

  5. [5]

    Comprehensive review on Alzheimer’s disease: causes and treatment,

    Z. Breijyeh, and R. Karaman, “Comprehensive review on Alzheimer’s disease: causes and treatment,” Molecules, vol. 25, no. 24, pp. 5789, 2020

    work page 2020

  6. [6]

    Genetic differences in the immediate transcriptome response to stress predict risk -related brain function and psychiatric disorders,

    J. Arloth, R. Bogdan, P. Weber et al. , “Genetic differences in the immediate transcriptome response to stress predict risk -related brain function and psychiatric disorders,” Neuron, vol. 86, no. 5, pp. 1189-1202, 2015

    work page 2015

  7. [7]

    MCAN: multimodal causal adversarial networks for dynamic effective connectivity learning from fMRI and EEG data,

    J. Liu, L. Han, and J. Ji, “MCAN: multimodal causal adversarial networks for dynamic effective connectivity learning from fMRI and EEG data,” IEEE Transactions on Medical Imaging, 2024

    work page 2024

  8. [8]

    Joint sparse representation of brain activity patterns in multi-task fMRI data,

    M. Ramezani, K. Marble, H. Trang et al., “Joint sparse representation of brain activity patterns in multi-task fMRI data,” IEEE transactions on medical imaging, vol. 34, no. 1, pp. 2-12, 2014

    work page 2014

Show all 54 references
  1. [9]

    Interpretable cognitive ability prediction: A comprehensive gated graph transformer framework for analyzing functional brain networks,

    G. Qu, A. Orlichenko, J. Wang et al., “Interpretable cognitive ability prediction: A comprehensive gated graph transformer framework for analyzing functional brain networks,” IEEE Transactions on Medical Imaging, 2023

    2023

  2. [10]

    Identifying autism spectrum disorder from resting-state fMRI using deep belief network,

    Z.-A. Huang, Z. Zhu, C. H. Yau et al., “Identifying autism spectrum disorder from resting-state fMRI using deep belief network,” IEEE Transactions on neural networks and learning systems, vol. 32, no. 7, pp. 2847-2861, 2020

    2020

  3. [11]

    GATE: Graph CCA for temporal self -supervised learning for label -efficient fMRI analysis,

    L. Peng, N. Wang, J. Xu et al. , “GATE: Graph CCA for temporal self -supervised learning for label -efficient fMRI analysis,” IEEE Transactions on Medical Imaging, vol. 42, no. 2, pp. 391-402, 2022

    2022

  4. [12]

    Modeling task fMRI data via deep convolutional autoencoder,

    H. Huang, X. Hu, Y. Zhao et al., “Modeling task fMRI data via deep convolutional autoencoder,” IEEE transactions on medical imaging, vol. 37, no. 7, pp. 1551 -1561, 2017

    2017

  5. [13]

    Federated multi-task learning for joint diagnosis of multiple mental disorders on MRI scans,

    Z.-A. Huang, Y. Hu, R. Liu et al., “Federated multi-task learning for joint diagnosis of multiple mental disorders on MRI scans,” IEEE Transactions on Biomedical Engineering, vol. 70, no. 4, pp. 1137-1149, 2022

    2022

  6. [14]

    Multi -hypergraph learning -based brain functional connectivity analysis in fMRI data,

    L. Xiao, J. Wang, P. H. Kassani et al. , “Multi -hypergraph learning -based brain functional connectivity analysis in fMRI data,” IEEE transactions on medical imaging, vol. 39, no. 5, pp. 1746-1758, 2019

    2019

  7. [15]

    Estimating effective connectivity by recurrent generative adversarial networks,

    J. Ji, J. Liu, L. Han et al., “Estimating effective connectivity by recurrent generative adversarial networks,” IEEE Transactions on Medical Imaging, vol. 40, no. 12, pp. 3326-3336, 2021

    2021

  8. [16]

    Multitask Learning for Joint Diagnosis of Multiple Mental Disorders in Resting-State fMRI,

    Z.-A. Huang, R. Liu, Z. Zhu et al., “Multitask Learning for Joint Diagnosis of Multiple Mental Disorders in Resting-State fMRI,” IEEE Transactions on Neural Networks and Learning Systems, 2022

    2022

  9. [17]

    Attention -like multimodality fusion with data augmentation for diagnosis of mental disorders using MRI,

    R. Liu, Z. -A. Huang, Y. Hu et al. , “Attention -like multimodality fusion with data augmentation for diagnosis of mental disorders using MRI,” IEEE Transactions on Neural Networks and Learning Systems, 2022. 29

    2022

  10. [18]

    Toward best practices for imaging transcriptomics of the human brain,

    A. Arnatkeviciute, R. D. Markello, B. D. Fulcher et al., “Toward best practices for imaging transcriptomics of the human brain,” Biological Psychiatry, vol. 93, no. 5, pp. 391-404, 2023

    2023

  11. [19]

    Spatial–temporal co-attention learning for diagnosis of mental disorders from resting -state fMRI data,

    R. Liu, Z.-A. Huang, Y. Hu et al., “Spatial–temporal co-attention learning for diagnosis of mental disorders from resting -state fMRI data,” IEEE Transactions on Neural Networks and Learning Systems, 2023

    2023

  12. [20]

    Source free semi-supervised transfer learning for diagnosis of mental disorders on fmri scans,

    Y. Hu, Z.-A. Huang, R. Liu et al., “Source free semi-supervised transfer learning for diagnosis of mental disorders on fmri scans,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    2023

  13. [21]

    A critical appraisal of imaging transcriptomics,

    A. S. Mandal, M. Gandal, J. Seidlitz et al. , “A critical appraisal of imaging transcriptomics,” Biological Psychiatry: Global Open Science, vol. 2, no. 4, pp. 311 - 313, 2022

    2022

  14. [22]

    The Allen Human Brain Atlas: comprehensive gene expression mapping of the human brain,

    E. H. Shen, C. C. Overly, and A. R. Jones, “The Allen Human Brain Atlas: comprehensive gene expression mapping of the human brain,” Trends in neurosciences, vol. 35, no. 12, pp. 711-714, 2012

    2012

  15. [23]

    Single-cell transcriptomic analysis of Alzheimer’s disease,

    H. Mathys, J. Davila-Velderrain, Z. Peng et al., “Single-cell transcriptomic analysis of Alzheimer’s disease,” Nature, vol. 570, no. 7761, pp. 332-337, 2019

    2019

  16. [24]

    A study of feature extraction for Alzheimer's disease based on resting-state fMRI

    S. Mao, C. Zhang, N. Gao et al., "A study of feature extraction for Alzheimer's disease based on resting-state fMRI." pp. 517-520

  17. [25]

    A supervised method to assist the diagnosis and monitor progression of Alzheimer's disease using data from an fMRI experiment,

    E. E. Tripoliti, D. I. Fotiadis, and M. Argyropoulou, “A supervised method to assist the diagnosis and monitor progression of Alzheimer's disease using data from an fMRI experiment,” Artificial intelligence in medicine, vol. 53, no. 1, pp. 35-45, 2011

    2011

  18. [26]

    Identification of the early stage of Alzheimer's disease using structural MRI and resting -state fMRI,

    S. H. Hojjati, A. Ebrahimzadeh, and A. Babajani -Feremi, “Identification of the early stage of Alzheimer's disease using structural MRI and resting -state fMRI,” Frontiers in neurology, vol. 10, pp. 904, 2019

    2019

  19. [27]

    Multi-scale time-series kernel-based learning method for brain disease diagnosis,

    Z. Zhang, J. Ding, J. Xu et al., “Multi-scale time-series kernel-based learning method for brain disease diagnosis,” IEEE journal of Biomedical and Health Informatics, vol. 25, no. 1, pp. 209-217, 2020

    2020

  20. [28]

    Analysis of Alzheimer’s disease based on the random neural network cluster in fMRI,

    X.-a. Bi, Q. Jiang, Q. Sun et al., “Analysis of Alzheimer’s disease based on the random neural network cluster in fMRI,” Frontiers in neuroinformatics, vol. 12, pp. 60, 2018

    2018

  21. [29]

    Improved brain age estimation with slice - based set networks

    U. Gupta, P. K. Lam, G. Ver Steeg et al., "Improved brain age estimation with slice - based set networks." pp. 840-844

  22. [30]

    Analysis of features of Alzheimer’s disease: Detection of early stage from functional brain changes in magnetic resonance images using a finetuned ResNet18 network,

    M. Odusami, R. Maskeliūnas, R. Damaševičius et al. , “Analysis of features of Alzheimer’s disease: Detection of early stage from functional brain changes in magnetic resonance images using a finetuned ResNet18 network,” Diagnostics, vol. 11, no. 6, pp. 1071, 2021

    2021

  23. [31]

    Multimodal data fusion of deep learning and dynamic functional connectivity features to predict Alzheimer’s disease progression

    A. Abrol, Z. Fu, Y. Du et al., "Multimodal data fusion of deep learning and dynamic functional connectivity features to predict Alzheimer’s disease progression." pp. 4409- 4413

  24. [32]

    Classification of Alzheimer’s Disease and Mild - Cognitive Impairment Base on High -Order Dynamic Functional Connectivity at Different Frequency Band,

    U. Khatri, and G. -R. Kwon, “Classification of Alzheimer’s Disease and Mild - Cognitive Impairment Base on High -Order Dynamic Functional Connectivity at Different Frequency Band,” Mathematics, vol. 10, no. 5, pp. 805, 2022

    2022

  25. [33]

    Mixture -of-experts with expert choice routing,

    Y. Zhou, T. Lei, H. Liu et al. , “Mixture -of-experts with expert choice routing,” Advances in Neural Information Processing Systems, vol. 35, pp. 7103-7114, 2022. 30

    2022

  26. [34]

    Protgnn: Towards self -explaining graph neural networks

    Z. Zhang, Q. Liu, H. Wang et al. , "Protgnn: Towards self -explaining graph neural networks." pp. 9127-9135

  27. [35]

    A survey of monte carlo tree search methods,

    C. B. Browne, E. Powley, D. Whitehouse et al., “A survey of monte carlo tree search methods,” IEEE Transactions on Computational Intelligence and AI in games, vol. 4, no. 1, pp. 1-43, 2012

    2012

  28. [36]

    PLINK: a tool set for whole -genome association and population-based linkage analyses,

    S. Purcell, B. Neale, K. Todd -Brown et al. , “PLINK: a tool set for whole -genome association and population-based linkage analyses,” The American journal of human genetics, vol. 81, no. 3, pp. 559-575, 2007

    2007

  29. [37]

    SCENIC: single -cell regulatory network inference and clustering,

    S. Aibar, C. B. Gonzá lez-Blas, T. Moerman et al., “SCENIC: single -cell regulatory network inference and clustering,” Nature methods, vol. 14, no. 11, pp. 1083 -1086, 2017

    2017

  30. [38]

    Integrated multimodal cell atlas of Alzheimer’s disease,

    M. I. Gabitto, K. J. Travaglini, V. M. Rachleff et al., “Integrated multimodal cell atlas of Alzheimer’s disease,” Research Square, 2023

    2023

  31. [39]

    The Alzheimer’s disease neuroimaging initiative,

    S. G. Mueller, M. W. Weiner, L. J. Thal et al., “The Alzheimer’s disease neuroimaging initiative,” Neuroimaging Clinics of North America, vol. 15, no. 4, pp. 869, 2005

    2005

  32. [40]

    The Alzheimer's disease neuroimaging initiative: progress report and future plans,

    M. W. Weiner, P. S. Aisen, C. R. Jack Jr et al., “The Alzheimer's disease neuroimaging initiative: progress report and future plans,” Alzheimer's & Dementia, vol. 6, no. 3, pp. 202-211. e7, 2010

    2010

  33. [41]

    Z. Chen, K. Chen, Y. Li et al. , “Structural, static, and dynamic functional MRI predictors for conversion from mild cognitive impairment to Alzheimer's disease: Inter‐cohort validation of Shanghai Memory Study and ADNI,” Human Brain Mapping, vol. 45, no. 1, pp. e26529, 2024

    2024

  34. [42]

    Comparison of the diagnostic accuracy of resting-state fMRI driven machine learning algorithms in the detection of mild cognitive impairment,

    G. Bolla, D. B. Berente, A. Andrá ssy et al., “Comparison of the diagnostic accuracy of resting-state fMRI driven machine learning algorithms in the detection of mild cognitive impairment,” Scientific Reports, vol. 13, no. 1, pp. 22285, 2023

    2023

  35. [43]

    The identification of Alzheimer’s disease using functional connectivity between activity voxels in resting -state fMRI data,

    Y. Shi, W. Zeng, J. Deng et al. , “The identification of Alzheimer’s disease using functional connectivity between activity voxels in resting -state fMRI data,” IEEE Journal of Translational Engineering in Health and Medicine, vol. 8, pp. 1-11, 2020

    2020

  36. [44]

    Gaussian process classification of Alzheimer's disease and mild cognitive impairment from resting-state fMRI,

    E. Challis, P. Hurley, L. Serra et al., “Gaussian process classification of Alzheimer's disease and mild cognitive impairment from resting-state fMRI,” NeuroImage, vol. 112, pp. 232-243, 2015

    2015

  37. [45]

    Classification of Alzheimer’s disease based on core -large scale brain network using multilayer extreme learning machine,

    R. K. Lama, J.-I. Kim, and G.-R. Kwon, “Classification of Alzheimer’s disease based on core -large scale brain network using multilayer extreme learning machine,” Mathematics, vol. 10, no. 12, pp. 1967, 2022

    1967

  38. [46]

    A. L. Martel, P. Abolmaesumi, D. Stoyanov et al. , Medical Image Computing and Computer Assisted Intervention–MICCAI 2020: 23rd International Conference, Lima, Peru, October 4–8, 2020, Proceedings, Part I: Springer Nature, 2020

    2020

  39. [47]

    Classification of brain disorders in rs -fMRI via local-to-global graph neural networks,

    H. Zhang, R. Song, L. Wang et al., “Classification of brain disorders in rs -fMRI via local-to-global graph neural networks,” IEEE transactions on medical imaging, vol. 42, no. 2, pp. 444-455, 2022

    2022

  40. [48]

    Remote sensing image classification based on a cross -attention mechanism and graph convolution,

    W. Cai, and Z. Wei, “Remote sensing image classification based on a cross -attention mechanism and graph convolution,” IEEE Geoscience and Remote Sensing Letters, vol. 19, pp. 1-5, 2020

    2020

  41. [49]

    Grad-CAM: Why did you say that?,

    R. R. Selvaraju, A. Das, R. Vedantam et al., “Grad-CAM: Why did you say that?,” 31 arXiv preprint arXiv:1611.07450, 2016

    2016 arXiv

  42. [50]

    Integration of multimodal data for deciphering brain disorders,

    J. Chen, G. Dong, L. Song et al., “Integration of multimodal data for deciphering brain disorders,” Annual Review of Biomedical Data Science, vol. 4, no. 1, pp. 43-56, 2021

    2021

  43. [51]

    Quantitative systems pharmacology in neuroscience: Novel methodologies and technologies,

    P. Bloomingdale, T. Karelina, M. Cirit et al., “Quantitative systems pharmacology in neuroscience: Novel methodologies and technologies,” CPT: Pharmacometrics & Systems Pharmacology, vol. 10, no. 5, pp. 412-419, 2021

    2021

  44. [52]

    Current status of tissue clearing and the path forward in neuroscience,

    J. Zhao, H. M. Lai, Y. Qi et al., “Current status of tissue clearing and the path forward in neuroscience,” ACS chemical neuroscience, vol. 12, no. 1, pp. 5-29, 2020

    2020

  45. [53]

    Cross-modal contrastive learning for text-to- image generation

    H. Zhang, J. Y. Koh, J. Baldridge et al., "Cross-modal contrastive learning for text-to- image generation." pp. 833-842

  46. [54]

    Unimo: Towards unified -modal understanding and generation via cross -modal contrastive learning,

    W. Li, C. Gao, G. Niu et al. , “Unimo: Towards unified -modal understanding and generation via cross -modal contrastive learning,” arXiv preprint arXiv:2012.15409 , 2020

    2012 arXiv

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.