pith. sign in

arxiv: 2605.24179 · v1 · pith:MQLDKKYTnew · submitted 2026-05-22 · 📡 eess.IV · q-bio.QM

7 Tesla Quantitative MRI and Machine Learning for Exploratory Motor Subtype Stratification and Diagnosis in Parkinson's Disease

Anne Louise Kristoffersen , Runa Geirmundsdatter Unsg{\aa}rd , Marc-Antoine Fortin , Ingrid Gylterud Kv{\aa}lsgard , Kjersti Eline Stige , Thanh Pierre Doan , Erik Magnus Berntsen , Charalampos Tzoulis
show 1 more author
P{\aa}l Erik Goa
This is my paper

Pith reviewed 2026-06-30 14:19 UTC · model grok-4.3

classification 📡 eess.IV q-bio.QM
keywords Parkinson's diseasequantitative MRImachine learningsubtype stratification7 Tesladeep learning segmentationmotor phenotypesfeature selection
0
0 comments X

The pith

Selected quantitative MRI features from 7T scans let machine learning separate Parkinson's motor subtypes with high accuracy in cross-validation.

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

The paper examines whether automatic segmentation of 7 Tesla quantitative MRI maps can yield brain-region features that, after selection, improve machine-learning classification of healthy controls versus Parkinson's patients and of the two main motor subtypes within patients. Three tasks were tested with 5-fold cross-validation on 21 controls and 24 patients: binary diagnosis, binary subtype separation, and three-class stratification. Using all features gave moderate performance, while selecting the best subset raised accuracies to 0.82, 1.00, and 0.73 respectively, indicating that compact imaging signatures may support objective subtype identification.

Core claim

Deep-learning U-Net segmentation of quantitative 7T MRI maps followed by feature selection produced classifiers whose performance exceeded that of models using every extracted feature, reaching perfect accuracy and AUC on the postural instability/gait difficulty versus tremor-dominant task and supporting the feasibility of low-dimensional, interpretable signatures for diagnosis support and phenotype stratification.

What carries the argument

Optimal subset selection performed on quantitative MRI values extracted from U-Net-segmented brain regions; the step reduces feature count and raises cross-validated accuracy on the three classification tasks.

If this is right

  • Low-dimensional imaging signatures become feasible for supporting objective motor-subtype assignment.
  • Feature selection after deep-learning segmentation demonstrably improves classification over use of all features.
  • Quantitative 7T maps combined with automatic segmentation can highlight differences between controls and the two motor phenotypes.
  • The approach opens a route toward imaging-supported study design and personalized treatment planning in heterogeneous Parkinson's disease.

Where Pith is reading between the lines

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

  • If the signatures prove stable, clinical rating scales could be supplemented or partially replaced by objective imaging metrics.
  • The same workflow might be tested on other movement disorders that also exhibit motor heterogeneity.
  • Scanner harmonization studies would be needed before multi-site deployment of the selected features.
  • Longitudinal scans could check whether the signatures track disease progression within each subtype.

Load-bearing premise

The 24 Parkinson's patients form a representative sample whose feature distributions will generalize, and feature selection on this cohort will not produce classifiers whose reported accuracies, especially the perfect score, will replicate on new data.

What would settle it

Running the same pipeline on an independent cohort of comparable size and finding that the selected-feature classifiers drop below 0.80 accuracy on the binary tasks would falsify the reported utility of the signatures.

read the original abstract

Parkinson's disease (PD) is a highly heterogeneous disease, including which motor symptoms are dominating. Imaging biomarkers that support subtype stratification could also improve biological understanding and study design, and enable personalized treatment strategies. This study evaluates whether deep-learning based automatic brain segmentation, in addition to quantitative maps from 7 Tesla MRI, can highlight differences between Healthy Controls (HC), Postural Instability and Gait Difficulty (PIGD) and Tremor Dominant (TD), and subsequently be used for objective PD stratification. The performance of machine learning classifiers may be improved with feature selection. 21 HC, and 24 people with PD (PwP) were included. The U-Net training was assessed with DSC. Two classification approaches using 5-fold cross-validation were defined across three tasks: (1) HC vs PwP; (2) PIGD vs TD; (3) multiclass, HC vs PIGD vs TD. Approach A used all extracted features. Approach B found the optimal subset of features for the classification tasks. The U-Net achieved mean DSC of 0.86 for all ROIs during training. Approach A: Task 1 best accuracy of 0.69 and best AUC of 0.73. Task 2 accuracy 0.69, AUC 0.90. Task 3 accuracy 0.62, AUC 0.66. Approach B: Task 1 accuracy of 0.82 and AUC of 0.93. Task 2 accuracy 1.00, AUC 1.00. Task 3 accuracy 0.73, AUC 0.91. DL-based segmentation combined with qMRI feature selection improved classification relative to using all features, supporting the potential of interpretable, low-dimensional imaging signatures for PD diagnosis support and phenotype stratification. Larger, multi-site studies are warranted to assess generalizability and stability.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 2 minor

Summary. The paper claims that deep-learning U-Net segmentation of 7T quantitative MRI maps, followed by machine-learning classification with feature selection (Approach B), improves performance over using all features (Approach A) across three tasks—HC vs PwP, PIGD vs TD, and multiclass—on a cohort of 21 HC and 24 PwP, yielding accuracies up to 0.82/1.00/0.73 and supporting low-dimensional interpretable imaging signatures for PD diagnosis and motor-subtype stratification.

Significance. If validated, the work would demonstrate a pipeline linking automated qMRI segmentation to phenotype stratification in a small PD cohort, potentially aiding biological subtyping; however, the reported gains rest on unverified feature-selection procedures whose stability on n=24 is unproven.

major comments (3)
  1. [Abstract / Methods (Approach B)] Abstract and Methods (Approach B description): Feature selection is described as finding 'the optimal subset of features for the classification tasks' prior to reporting 5-fold CV results, with no statement that selection occurs inside each training fold. On a total of 24 PD patients this procedure risks selecting sample-specific features, directly undermining the claim that Approach B improves classification (e.g., Task 2 accuracy rising from 0.69 to 1.00).
  2. [Results (classification performance)] Results (Task 2, PIGD vs TD): The reported accuracy of 1.00 and AUC of 1.00 after feature selection on only 24 patients is statistically implausible without external validation or nested cross-validation; this single result is load-bearing for the central claim of 'interpretable, low-dimensional imaging signatures.'
  3. [Methods (classification approaches)] Methods (cross-validation details): The manuscript provides no description of nested cross-validation, held-out test set, or baseline clinical classifiers, leaving the reported gains (Approach B vs A) without a control for selection bias on small N.
minor comments (2)
  1. [Abstract] Abstract: The sentence 'The U-Net achieved mean DSC of 0.86 for all ROIs during training' should clarify whether this is training or validation DSC and list the ROIs.
  2. [Methods] The total sample size (N=45) and the split between PIGD and TD subgroups should be stated explicitly in the Methods when describing the classification tasks.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments highlighting important methodological concerns with our small-cohort study. We address each major comment below and commit to revisions that improve transparency and rigor without overstating the current results.

read point-by-point responses

  1. Referee: [Abstract / Methods (Approach B)] Abstract and Methods (Approach B description): Feature selection is described as finding 'the optimal subset of features for the classification tasks' prior to reporting 5-fold CV results, with no statement that selection occurs inside each training fold. On a total of 24 PD patients this procedure risks selecting sample-specific features, directly undermining the claim that Approach B improves classification (e.g., Task 2 accuracy rising from 0.69 to 1.00).

    Authors: We agree that the manuscript description is ambiguous and does not confirm feature selection occurred inside the CV folds. This is a valid concern for selection bias on n=24. In revision we will rewrite the Methods to specify nested cross-validation, with feature selection performed independently on each training fold only, and will recompute and report the revised performance metrics. revision: yes

  2. Referee: [Results (classification performance)] Results (Task 2, PIGD vs TD): The reported accuracy of 1.00 and AUC of 1.00 after feature selection on only 24 patients is statistically implausible without external validation or nested cross-validation; this single result is load-bearing for the central claim of 'interpretable, low-dimensional imaging signatures.'

    Authors: We acknowledge that perfect separation on this small subsample is likely inflated by the non-nested feature selection and should not be presented as robust evidence. In the revision we will replace the current Task 2 numbers with results from properly nested CV, add explicit caveats in Results and Discussion about small-sample instability, and soften the language around the signatures to reflect that they are exploratory. revision: yes

  3. Referee: [Methods (classification approaches)] Methods (cross-validation details): The manuscript provides no description of nested cross-validation, held-out test set, or baseline clinical classifiers, leaving the reported gains (Approach B vs A) without a control for selection bias on small N.

    Authors: We will expand the Methods to detail the nested CV procedure and state that a held-out test set was not used because of the limited total sample (n=45). We did not compare against clinical classifiers because the study focus was imaging-derived features; a brief note on this scope limitation will be added, but a full clinical baseline comparison is outside the current scope. revision: partial

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents empirical ML classification results (Approach A vs B) obtained via 5-fold cross-validation on a small cohort, with no equations, self-citations, or derivation chain that reduces a claimed result to its inputs by construction. Feature selection is described as one of two defined approaches within the CV tasks, but the text supplies no explicit statement that selection occurs outside the CV loop or that performance metrics are forced by the selection step itself. No load-bearing self-citation, ansatz smuggling, or renaming of known results appears. The reported accuracies are therefore treated as direct experimental outcomes rather than tautological outputs.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Central claim depends on unstated assumptions that qMRI values in segmented ROIs capture biologically stable subtype differences and that the 5-fold CV with feature selection yields generalizable signatures; no independent evidence for either is supplied in the abstract.

free parameters (1)
  • feature selection criterion and subset size
    Optimal subset chosen on the same data; exact method and number of retained features not stated.
axioms (1)
  • domain assumption U-Net segmentation produces ROIs whose quantitative values are reliable across subjects
    Mean DSC 0.86 is reported but no per-ROI or inter-subject variability analysis is given.

pith-pipeline@v0.9.1-grok · 5949 in / 1391 out tokens · 41453 ms · 2026-06-30T14:19:56.434891+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

47 extracted references · 14 canonical work pages · 1 internal anchor

  1. [1]

    Clinical and Pathological Features of Parkinson’s Disease

    Schneider SA, Obeso JA. Clinical and Pathological Features of Parkinson’s Disease. In: Nguyen HHP, Cenci MA, editors. Behavioral Neurobiology of Huntington’s Disease and Parkinson’s Disease. Berlin, Heidelberg: Springer; 2015. p. 205–220. Availablefrom:https://doi.org/10.1007/7854_2014_317

    work page doi:10.1007/7854_2014_317 2015

  2. [2]

    Parkinson’s disease

    Lees AJ, Hardy J, Revesz T. Parkinson’s disease. The Lancet. 2009 Jun;373(9680):2055–2066. Available from: https://www. sciencedirect.com/science/article/pii/S014067360960492X

    2009

  3. [3]

    The Emerging Evidence of the Parkinson Pandemic

    Dorsey ER, Sherer T, Okun MS, Bloem BR. The Emerging Evidence of the Parkinson Pandemic. Journal of Parkinson’s Disease.2018Dec;8(s1):S3–S8. Availablefrom:https://doi.org/10.3233/JPD-181474

    work page doi:10.3233/jpd-181474

  4. [4]

    TheclinicalheterogeneityofParkinson’sdiseaseanditstherapeuticimplica- tions

    GreenlandJC,Williams-GrayCH,BarkerRA. TheclinicalheterogeneityofParkinson’sdiseaseanditstherapeuticimplica- tions. European Journal of Neuroscience. 2019;49(3):328–338. Available from: https://onlinelibrary.wiley.com/doi/abs/10. 1111/ejn.14094

    2019

  5. [5]

    Finding useful biomarkers for Parkinson’s disease.ScienceTranslationalMedicine.2018Aug;10(454):eaam6003.Availablefrom:https://www.science.org/doi/10.1126/ scitranslmed.aam6003

    Chen-Plotkin AS, Albin R, Alcalay R, Babcock D, Bajaj V, Bowman D, et al. Finding useful biomarkers for Parkinson’s disease.ScienceTranslationalMedicine.2018Aug;10(454):eaam6003.Availablefrom:https://www.science.org/doi/10.1126/ scitranslmed.aam6003

  6. [6]

    Significance of MRI in Diagnosis and Differential Diagnosis of Parkinson’s Disease

    Mahlknecht P, Hotter A, Hussl A, Esterhammer R, Schocke M, Seppi K. Significance of MRI in Diagnosis and Differential Diagnosis of Parkinson’s Disease. Neurodegenerative Diseases. 2010 Jul;7(5):300–318. Available from: https://doi.org/10. 1159/000314495

    2010

  7. [7]

    Magnetic resonance imaging for the diagnosis of Parkinson’s disease

    Heim B, Krismer F, De Marzi R, Seppi K. Magnetic resonance imaging for the diagnosis of Parkinson’s disease. Journal of NeuralTransmission.2017;124(8):915–964. Availablefrom:https://pmc.ncbi.nlm.nih.gov/articles/PMC5514207/

    2017

  8. [8]

    Parkinson’s disease

    Bloem BR, Okun MS, Klein C. Parkinson’s disease. The Lancet. 2021 Jun;397(10291):2284–2303. Available from: https: //www.sciencedirect.com/science/article/pii/S014067362100218X

    2021

  9. [9]

    ParkinsonDiseaseSubtypes

    ThenganattMA,JankovicJ. ParkinsonDiseaseSubtypes. JAMANeurology.2014Apr;71(4):499–504. Availablefrom:https: //doi.org/10.1001/jamaneurol.2013.6233

    work page doi:10.1001/jamaneurol.2013.6233 2013

  10. [10]

    Parkinson’s disease

    Kalia LV, Lang AE. Parkinson’s disease. The Lancet. 2015 Aug;386(9996):896–912. Available from: https://www. sciencedirect.com/science/article/pii/S0140673614613933

    2015

  11. [11]

    EnhancingParkinson’sDiseaseDiagnosisthroughDeepLearning-BasedClas- sificationof3DMRIImages

    DesaiS,ChhinkaniwalaH,ShahS,GajjarP. EnhancingParkinson’sDiseaseDiagnosisthroughDeepLearning-BasedClas- sificationof3DMRIImages. ProcediaComputerScience.2024Jan;235:201–213. Availablefrom:https://www.sciencedirect. com/science/article/pii/S1877050924006999

  12. [12]

    ClassificationofParkinson’sdiseaseusing3DConvolutionalNeuralNetworks (CNN)

    ZubairM,FerranteM,DelGrattaC,ToschiN. ClassificationofParkinson’sdiseaseusing3DConvolutionalNeuralNetworks (CNN). Marseille, France; 2024. Available from: https://www.esmrmb2024.org/abstracts-form/posters-e/abstract-data/ 5c3ac5bc0f8b303c68cf6a912717c02f

    2024

  13. [13]

    SyntheticMRIstudyofbrainvolumeandsubcorticalmyelininvarious Parkinson’sdiseasemotorsubtypes

    ChengD,WenJ,DingN,DuanZ,LinB,YangY,etal. SyntheticMRIstudyofbrainvolumeandsubcorticalmyelininvarious Parkinson’sdiseasemotorsubtypes. npjParkinson’sDisease.2025Aug;11(1):255. Availablefrom:https://www.nature.com/ articles/s41531-025-01120-x

  14. [14]

    Distributionpatternofirondepositioninthebasalgangliaofdifferentmotor subtypes of Parkinson’s disease

    ZhangX,LiL,QiL,FuY,SunD,ChenS,etal. Distributionpatternofirondepositioninthebasalgangliaofdifferentmotor subtypes of Parkinson’s disease. Neuroscience Letters. 2023 Jun;807:137249. Available from: https://www.sciencedirect. com/science/article/pii/S0304394023002082. 16of24

    2023

  15. [15]

    Quantitative susceptibility mapping as an indicator of subcortical and limbic iron abnormality in Parkinson’s disease with dementia

    Li DTH, Hui ES, Chan Q, Yao N, Chua SE, McAlonan GM, et al. Quantitative susceptibility mapping as an indicator of subcortical and limbic iron abnormality in Parkinson’s disease with dementia. NeuroImage: Clinical. 2018 Jan;20:365–373. Availablefrom:https://www.sciencedirect.com/science/article/pii/S2213158218302407

    2018

  16. [16]

    Brainirondepositionislinkedwithcognitive severity in Parkinson’s disease

    ThomasGEC,LeylandLA,SchragAE,LeesAJ,Acosta-CabroneroJ,WeilRS. Brainirondepositionislinkedwithcognitive severity in Parkinson’s disease. Journal of Neurology, Neurosurgery & Psychiatry. 2020 Apr;91(4):418–425. Available from: https://jnnp.bmj.com/content/91/4/418

    2020

  17. [17]

    Availablefrom:https://doi.org/10.3233/JPD-150729

    GuQ,ZhangH,XuanM,LuoW,HuangP,XiaS,etal.AutomaticClassificationonMulti-ModalMRIDataforDiagnosisofthe PosturalInstabilityandGaitDifficultySubtypeofParkinson’sDisease.JournalofParkinson’sDisease.2016Jun;6(3):545–556. Availablefrom:https://doi.org/10.3233/JPD-150729

    work page doi:10.3233/jpd-150729

  18. [18]

    Cross-regional radiomics: a novel framework for relationship-based feature ex- traction with validation in Parkinson’s disease motor subtyping

    Hosseini MS, Aghamiri SMR, Panahi M. Cross-regional radiomics: a novel framework for relationship-based feature ex- traction with validation in Parkinson’s disease motor subtyping. BioData Mining. 2025 Sep;18(1):67. Available from: https://doi.org/10.1186/s13040-025-00483-4

    work page doi:10.1186/s13040-025-00483-4 2025

  19. [19]

    ExplainableclassificationofParkinson’sdiseasewithdifferentmotor subtypesbyanalyzingthesyntheticMRIquantitativeparametersofsubcorticalnuclei

    ChengD,WenJ,LiuY,DingN,DuanZ,YangY,etal. ExplainableclassificationofParkinson’sdiseasewithdifferentmotor subtypesbyanalyzingthesyntheticMRIquantitativeparametersofsubcorticalnuclei. EuropeanJournalofRadiology.2025 Sep;190. Availablefrom:https://www.ejradiology.com/article/S0720-048X%2825%2900358-4/fulltext

    2025

  20. [20]

    TheSTRAT-PARKcohort:Apersonalizedinitiative to stratify Parkinson’s disease

    StigeKE,KvernengSU,SharmaS,SkeieGO,SheardE,SøgnenM,etal. TheSTRAT-PARKcohort:Apersonalizedinitiative to stratify Parkinson’s disease. Progress in Neurobiology. 2024 May;236:102603. Available from: https://www.sciencedirect. com/science/article/pii/S030100822400039X

    2024

  21. [21]

    Stebbins GT, Goetz CG, Burn DJ, Jankovic J, Khoo TK, Tilley BC. How to identify tremor dominant and postural instabili- ty/gait difficulty groups with the movement disorder society unified Parkinson’s disease rating scale: Comparison with the unified Parkinson’s disease rating scale. Movement Disorders. 2013;28(5):668–670. Available from: https://onlineli...

    work page doi:10.1002/mds.25383 2013

  22. [22]

    MP2RAGE, a self bias-field corrected sequenceforimprovedsegmentationandT1-mappingathighfield

    Marques JP, Kober T, Krueger G, van der Zwaag W, Van de Moortele PF, Gruetter R. MP2RAGE, a self bias-field corrected sequenceforimprovedsegmentationandT1-mappingathighfield. NeuroImage.2010Jan;49(2):1271–1281

  23. [23]

    ComputationallyEfficientCombinationofMulti- channelPhaseDataFromMulti-echoAcquisitions(ASPIRE)

    EcksteinK,DymerskaB,BachrataB,BognerW,PoljancK,TrattnigS,etal. ComputationallyEfficientCombinationofMulti- channelPhaseDataFromMulti-echoAcquisitions(ASPIRE). MagneticResonanceinMedicine.2018Jun;79(6):2996–3006

  24. [24]

    Fastquantitativesusceptibilitymappingusing3D EPIandtotalgeneralizedvariation

    LangkammerC,BrediesK,PoserBA,BarthM,ReishoferG,FanAP,etal. Fastquantitativesusceptibilitymappingusing3D EPIandtotalgeneralizedvariation. NeuroImage.2015May;111:622–630

  25. [25]

    Nighres: processing tools for high-resolution neuroimaging

    Huntenburg JM, Steele CJ, Bazin PL. Nighres: processing tools for high-resolution neuroimaging. GigaScience. 2018 Jul;7(7):giy082. Availablefrom:https://doi.org/10.1093/gigascience/giy082

    work page doi:10.1093/gigascience/giy082 2018

  26. [26]

    FastSurfer - A fast and accurate deep learning based neu- roimagingpipeline

    Henschel L, Conjeti S, Estrada S, Diers K, Fischl B, Reuter M. FastSurfer - A fast and accurate deep learning based neu- roimagingpipeline. NeuroImage.2020Oct;219:117012. Availablefrom:https://www.sciencedirect.com/science/article/pii/ S1053811920304985

  27. [27]

    FastSurferVINN:Buildingresolution-independenceintodeeplearningsegmentationmeth- ods—A solution for HighRes brain MRI

    HenschelL,KüglerD,ReuterM. FastSurferVINN:Buildingresolution-independenceintodeeplearningsegmentationmeth- ods—A solution for HighRes brain MRI. NeuroImage. 2022 May;251:118933. Available from: https://www.sciencedirect. com/science/article/pii/S1053811922000623

    2022

  28. [28]

    scikit-image:imageprocessinginPython

    WaltSvd,SchönbergerJL,Nunez-IglesiasJ,BoulogneF,WarnerJD,YagerN,etal. scikit-image:imageprocessinginPython. PeerJ.2014Jun;2:e453. Availablefrom:https://peerj.com/articles/453

  29. [29]

    nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation

    Isensee F, Jaeger PF, Kohl SAA, Petersen J, Maier-Hein KH. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature Methods. 2021 Feb;18(2):203–211. Available from: https://www.nature.com/ articles/s41592-020-01008-z

    2021

  30. [30]

    SPM12Software-StatisticalParametricMappingAvailablefrom:https://www.fil.ion.ucl.ac.uk/spm/software/spm12/

  31. [31]

    Availablefrom:http://jmlr.org/papers/v12/pedregosa11a.html

    PedregosaF,VaroquauxG,GramfortA,MichelV,ThirionB,GriselO,etal.Scikit-learn:MachineLearninginPython.Journal ofMachineLearningResearch.2011;12(85):2825–2830. Availablefrom:http://jmlr.org/papers/v12/pedregosa11a.html

    2011

  32. [32]

    XGBoost: A Scalable Tree Boosting System

    Chen T, Guestrin C. XGBoost: A Scalable Tree Boosting System. In: Proceedings of the 22nd ACM SIGKDD International ConferenceonKnowledgeDiscoveryandDataMining;2016.p.785–794. Availablefrom:http://arxiv.org/abs/1603.02754

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  33. [33]

    Mining imaging and clinical data with machine learning approaches for the diagnosis and early detection of Parkinson’s disease

    Zhang J. Mining imaging and clinical data with machine learning approaches for the diagnosis and early detection of Parkinson’s disease. npj Parkinson’s Disease. 2022 Jan;8(1):13. Available from: https://www.nature.com/articles/ s41531-021-00266-8. 17of24

    2022

  34. [34]

    ATutorialonSupportVectorMachinesforPatternRecognition

    BurgesCJC. ATutorialonSupportVectorMachinesforPatternRecognition. DataMiningandKnowledgeDiscovery.1998 Jun;2(2):121–167. Availablefrom:https://doi.org/10.1023/A:1009715923555

    work page doi:10.1023/a:1009715923555 1998

  35. [35]

    The parameter sensitivity of random forests

    Huang BFF, Boutros PC. The parameter sensitivity of random forests. BMC Bioinformatics. 2016 Sep;17(1):331. Available from:https://doi.org/10.1186/s12859-016-1228-x

    work page doi:10.1186/s12859-016-1228-x 2016

  36. [36]

    Digitalmedicineandthecurseofdimensionality

    BerishaV,KrantsevichC,HahnPR,HahnS,DasarathyG,TuragaP,etal. Digitalmedicineandthecurseofdimensionality. npjDigitalMedicine.2021Oct;4(1):153. Availablefrom:https://www.nature.com/articles/s41746-021-00521-5

  37. [37]

    Quantitative susceptibility mapping in Parkinson’s disease

    Langkammer C, Pirpamer L, Seiler S, Deistung A, Schweser F, Franthal S, et al. Quantitative susceptibility mapping in Parkinson’s disease. PLoS ONE. 2016;11(9). Available from: https://www.scopus.com/inward/record.uri?eid=2-s2. 0-84992374829&doi=10.1371%2fjournal.pone.0162460&partnerID=40&md5=a27c987b14555d0d2d473945daebae43

    2016

  38. [38]

    Regionally progressive accumulation of iron in Parkinson’s disease as measured by quantitative susceptibility mapping

    Guan X, Xuan M, Gu Q, Huang P, Liu C, Wang N, et al. Regionally progressive accumulation of iron in Parkinson’s disease as measured by quantitative susceptibility mapping. NMR in Biomedicine. 2017;30(4). Avail- able from: https://www.scopus.com/inward/record.uri?eid=2-s2.0-84975702009&doi=10.1002%2fnbm.3489&partnerID= 40&md5=1725c71633fdfdfdf110be1879bcd776

    2017

  39. [39]

    Utility of quantitative susceptibility map- ping and diffusion kurtosis imaging in the diagnosis of early Parkinson’s disease

    Tan S, Hartono S, Welton T, Ann CN, Lim SL, Koh TS, et al. Utility of quantitative susceptibility map- ping and diffusion kurtosis imaging in the diagnosis of early Parkinson’s disease. NeuroImage: Clinical. 2021;32. Available from: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85116354295&doi=10.1016%2fj.nicl.2021.102831& partnerID=40&md5=7e55227d7...

    work page arXiv 2021

  40. [40]

    Different iron distribution patterns in Parkinson’s disease and its motor subtypes: a quantitative susceptibility mapping study

    Zang S, Pan Y, Chen M, Zhang G. Different iron distribution patterns in Parkinson’s disease and its motor subtypes: a quantitative susceptibility mapping study. Acta Radiologica. 2025;66(1):99–106. Available from: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85209914284&doi=10.1177%2f02841851241297207&partnerID= 40&md5=9eabf33fd0b71453882452fc805df2f1

    2025

  41. [41]

    Evaluation of iron deposition in brain basal ganglia of patients with Parkinson’s disease using quantitative susceptibility mapping

    Shahmaei V, Faeghi F, Mohammdbeigi A, Hashemi H, Ashrafi F. Evaluation of iron deposition in brain basal ganglia of patients with Parkinson’s disease using quantitative susceptibility mapping. European Journal of Radiology Open. 2019;6:169–174. Available from: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85064746206&doi=10.1016%2fj. ejro.2019.04.00...

    2019

  42. [42]

    Region-specific disturbed iron distribution in early idiopathic Parkinson’sdiseasemeasuredbyquantitativesusceptibilitymapping

    He N, Ling H, Ding B, Huang J, Zhang Y, Zhang Z, et al. Region-specific disturbed iron distribution in early idiopathic Parkinson’sdiseasemeasuredbyquantitativesusceptibilitymapping. HumanBrainMapping.2015;36(11):4407–4420. Avail- able from: https://www.scopus.com/inward/record.uri?eid=2-s2.0-84955726895&doi=10.1002%2fhbm.22928&partnerID= 40&md5=fbf6e6285...

    2015

  43. [43]

    Quantifying brain iron deposition in patients with Parkinson’s disease using quantitative susceptibility mapping, R2 and R2*

    Barbosa JHO, Santos AC, Tumas V, Liu M, Zheng W, Haacke EM, et al. Quantifying brain iron deposition in patients with Parkinson’s disease using quantitative susceptibility mapping, R2 and R2*. Magnetic Resonance Imaging. 2015 Jun;33(5):559–565. Availablefrom:https://www.sciencedirect.com/science/article/pii/S0730725X15000600

    2015

  44. [44]

    Midbrain and pallidal iron changes identify patientswithREMsleepbehaviourdisorderandParkinson’sdisease

    Alushaj E, Kuurstra A, Menon RS, Ganjavi H, Morava A, Sharma M, et al. Midbrain and pallidal iron changes identify patientswithREMsleepbehaviourdisorderandParkinson’sdisease. npjParkinson’sDisease.2025Apr;11(1):84. Available from:https://www.nature.com/articles/s41531-025-00916-1

  45. [45]

    Network Modeling Analysis in Health Informatics and Bioinformatics

    AhmedYK,NajiANA.Smartfeatureextractionusingdeeplearningforearlydiagnosisofchronicdiseasesinnext-generation medical decision support systems. Network Modeling Analysis in Health Informatics and Bioinformatics. 2025;14(1):140. Availablefrom:https://doi.org/10.1007/s13721-025-00641-y

    work page doi:10.1007/s13721-025-00641-y 2025

  46. [46]

    FeatureExtractionBasedonDeepLearningforSomeTraditionalMachineLearningMethods

    ÇayirA,YenidoğanI,DağH. FeatureExtractionBasedonDeepLearningforSomeTraditionalMachineLearningMethods. In: 2018 3rd International Conference on Computer Science and Engineering (UBMK); 2018. p. 494–497. Available from: https://ieeexplore.ieee.org/abstract/document/8566383

    work page arXiv 2018

  47. [47]

    Swallow Tail Sign: Revisited

    Brammerloh M, Kirilina E, Alkemade A, Bazin PL, Jantzen C, Jäger C, et al. Swallow Tail Sign: Revisited. Radiology. 2022 Dec;305(3):674–677. Availablefrom:https://pubs.rsna.org/doi/full/10.1148/radiol.212696. 18of24 SupplementaryMaterial S.1 WorkflowforApproachB TheworkflowforobtainingthefeaturesforApproachB,optimalsubset,isshowninFigureS1forTask1:Healthy...

    work page doi:10.1148/radiol.212696 2022