Home > In process > Enhancing interpretability of AI with radiomics-based deep neural network: proof of concept in the classification of Parkinsonian syndromes with 18F-FDG PET imaging. > print

001     285004
005     20260202113140.0
024 7 _ |a 10.1007/s00259-025-07478-7
|2 doi
024 7 _ |a pmid:40932612
|2 pmid
024 7 _ |a 1619-7070
|2 ISSN
024 7 _ |a 1619-7089
|2 ISSN
037 _ _ |a DZNE-2026-00138
041 _ _ |a English
082 _ _ |a 610
100 1 _ |a Li, Chenyang
|b 0
245 _ _ |a Enhancing interpretability of AI with radiomics-based deep neural network: proof of concept in the classification of Parkinsonian syndromes with 18F-FDG PET imaging.
260 _ _ |a Heidelberg [u.a.]
|c 2026
|b Springer-Verl.
336 7 _ |a article
|2 DRIVER
336 7 _ |a Output Types/Journal article
|2 DataCite
336 7 _ |a Journal Article
|b journal
|m journal
|0 PUB:(DE-HGF)16
|s 1770028161_15530
|2 PUB:(DE-HGF)
336 7 _ |a ARTICLE
|2 BibTeX
336 7 _ |a JOURNAL_ARTICLE
|2 ORCID
336 7 _ |a Journal Article
|0 0
|2 EndNote
520 _ _ |a Interpretability and reproducibility remain major challenges in applying deep neural network (DNN) to neuroimaging-based diagnosis. This study proposes a radiomics-guided dual-channel deep neural network (RDDNN) to improve feature transparency and enhance clinical understanding in the classification of Parkinsonian syndromes.In this bi-centric study, we analysed two independent cohorts comprising 1,275 patients with idiopathic Parkinson's disease (IPD), multiple system atrophy (MSA), and progressive supranuclear palsy (PSP), alongside 223 healthy controls from Huashan Hospital and 90 patients with IPD, MSA, and PSP (34IPD, 17MSA, 39PSP) from the University Hospital Munich. It is a re-analysis of well-studied Chinese and German cohorts of 18F-fluorodeoxyglucose Positron emission tomography (FDG-PET) imaging of parkinsonian patients and the FDG scans were of 10-min static acquisition at 60 min post FDG injection and normalized against whole brain activity. The RDDNN model combines local features extracted via dilated convolutional networks and global features derived from Transformer-based self-attention networks. Model performance was evaluated using classification metrics and compared to radiomics and DNN approaches. The model's outputs were also compared with nuclear medicine specialists' visual assessments to assess interpretability and time efficiency. Furthermore, SHapley Additive Explanations (SHAP), Layer-wise Class Activation Mapping (Layer-CAM), and Rollout Attention Map (RAM) were employed to evaluate which features played the most critical roles in the model's final classification decisions after supervised training, and to examine how both networks spatially corresponded to known brain connectivity regions.In the internal blind-test cohort, the RDDNN achieved high accuracy (AUC = 0.99, accuracy = 0.98). SHAP and correlation analyses jointly indicated complementary information across channels, some of which were clinically interpretable. In the external cohort, the model maintained robust performance (AUC = 0.94, accuracy = 0.81), with consistent feature patterns across populations. The model significantly reduced evaluation time compared to nuclear medicine specialists' readings (p < 0.001), and the heatmaps showed disease-specific activation in anatomically relevant regions for IPD, MSA, and PSP.The RDDNN framework provides a clinically interpretable and reproducible DNN solution for classifying Parkinsonian disorders. By integrating radiomics and attention-based modeling, it enhances lesion localization, supports clinical decision-making, and offers performance comparable to human specialists-while substantially improving diagnostic efficiency.
536 _ _ |a 353 - Clinical and Health Care Research (POF4-353)
|0 G:(DE-HGF)POF4-353
|c POF4-353
|f POF IV
|x 0
536 _ _ |a 352 - Disease Mechanisms (POF4-352)
|0 G:(DE-HGF)POF4-352
|c POF4-352
|f POF IV
|x 1
588 _ _ |a Dataset connected to CrossRef, PubMed, , Journals: pub.dzne.de
650 _ 7 |a 18F-fluorodeoxyglucosePET imaging
|2 Other
650 _ 7 |a Dual-channel neural network
|2 Other
650 _ 7 |a Model interpretability
|2 Other
650 _ 7 |a Parkinsonian syndromes
|2 Other
650 _ 7 |a Radiomics
|2 Other
650 _ 7 |a SHAP
|2 Other
650 _ 7 |a Fluorodeoxyglucose F18
|0 0Z5B2CJX4D
|2 NLM Chemicals
650 _ 2 |a Humans
|2 MeSH
650 _ 2 |a Fluorodeoxyglucose F18
|2 MeSH
650 _ 2 |a Parkinsonian Disorders: diagnostic imaging
|2 MeSH
650 _ 2 |a Parkinsonian Disorders: classification
|2 MeSH
650 _ 2 |a Positron-Emission Tomography
|2 MeSH
650 _ 2 |a Male
|2 MeSH
650 _ 2 |a Female
|2 MeSH
650 _ 2 |a Neural Networks, Computer
|2 MeSH
650 _ 2 |a Middle Aged
|2 MeSH
650 _ 2 |a Aged
|2 MeSH
650 _ 2 |a Proof of Concept Study
|2 MeSH
650 _ 2 |a Deep Learning
|2 MeSH
650 _ 2 |a Image Processing, Computer-Assisted: methods
|2 MeSH
650 _ 2 |a Radiomics
|2 MeSH
700 1 _ |a Jiao, Fangyang
|b 1
700 1 _ |a Wu, Shaoyou
|b 2
700 1 _ |a Wang, Chenhan
|b 3
700 1 _ |a Wei, Min
|b 4
700 1 _ |a Zhang, Shuoyan
|b 5
700 1 _ |a Wang, Luyao
|b 6
700 1 _ |a Huang, Yu
|b 7
700 1 _ |a Yin, Yafu
|b 8
700 1 _ |a Tian, Rong
|b 9
700 1 _ |a Bernhardt, Alexander
|0 P:(DE-2719)9002620
|b 10
|u dzne
700 1 _ |a Katzdobler, Sabrina
|0 P:(DE-2719)9001160
|b 11
|u dzne
700 1 _ |a Levin, Johannes
|0 P:(DE-2719)2811659
|b 12
|u dzne
700 1 _ |a Höglinger, Günter U
|0 P:(DE-2719)2811373
|b 13
|u dzne
700 1 _ |a Brendel, Matthias
|0 P:(DE-2719)9001539
|b 14
|u dzne
700 1 _ |a Rominger, Axel
|b 15
700 1 _ |a Shi, Kuangyu
|b 16
700 1 _ |a Zuo, Chuantao
|b 17
700 1 _ |a Jiang, Jiehui
|0 0000-0003-4948-3683
|b 18
773 _ _ |a 10.1007/s00259-025-07478-7
|g Vol. 53, no. 3, p. 1962 - 1979
|0 PERI:(DE-600)2098375-X
|n 3
|p 1962 - 1979
|t European journal of nuclear medicine and molecular imaging
|v 53
|y 2026
|x 1619-7070
856 4 _ |u https://pub.dzne.de/record/285004/files/DZNE-2026-00138.pdf
|y Restricted
856 4 _ |u https://pub.dzne.de/record/285004/files/DZNE-2026-00138.pdf?subformat=pdfa
|x pdfa
|y Restricted
910 1 _ |a Deutsches Zentrum für Neurodegenerative Erkrankungen
|0 I:(DE-588)1065079516
|k DZNE
|b 10
|6 P:(DE-2719)9002620
910 1 _ |a Deutsches Zentrum für Neurodegenerative Erkrankungen
|0 I:(DE-588)1065079516
|k DZNE
|b 11
|6 P:(DE-2719)9001160
910 1 _ |a Deutsches Zentrum für Neurodegenerative Erkrankungen
|0 I:(DE-588)1065079516
|k DZNE
|b 12
|6 P:(DE-2719)2811659
910 1 _ |a Deutsches Zentrum für Neurodegenerative Erkrankungen
|0 I:(DE-588)1065079516
|k DZNE
|b 13
|6 P:(DE-2719)2811373
910 1 _ |a Deutsches Zentrum für Neurodegenerative Erkrankungen
|0 I:(DE-588)1065079516
|k DZNE
|b 14
|6 P:(DE-2719)9001539
913 1 _ |a DE-HGF
|b Gesundheit
|l Neurodegenerative Diseases
|1 G:(DE-HGF)POF4-350
|0 G:(DE-HGF)POF4-353
|3 G:(DE-HGF)POF4
|2 G:(DE-HGF)POF4-300
|4 G:(DE-HGF)POF
|v Clinical and Health Care Research
|x 0
913 1 _ |a DE-HGF
|b Gesundheit
|l Neurodegenerative Diseases
|1 G:(DE-HGF)POF4-350
|0 G:(DE-HGF)POF4-352
|3 G:(DE-HGF)POF4
|2 G:(DE-HGF)POF4-300
|4 G:(DE-HGF)POF
|v Disease Mechanisms
|x 1
915 _ _ |a DEAL Springer
|0 StatID:(DE-HGF)3002
|2 StatID
|d 2025-11-07
|w ger
915 _ _ |a DEAL Springer
|0 StatID:(DE-HGF)3002
|2 StatID
|d 2025-11-07
|w ger
915 _ _ |a DBCoverage
|0 StatID:(DE-HGF)0200
|2 StatID
|b SCOPUS
|d 2025-11-07
915 _ _ |a DBCoverage
|0 StatID:(DE-HGF)0300
|2 StatID
|b Medline
|d 2025-11-07
915 _ _ |a DBCoverage
|0 StatID:(DE-HGF)0199
|2 StatID
|b Clarivate Analytics Master Journal List
|d 2025-11-07
915 _ _ |a DBCoverage
|0 StatID:(DE-HGF)1110
|2 StatID
|b Current Contents - Clinical Medicine
|d 2025-11-07
915 _ _ |a DBCoverage
|0 StatID:(DE-HGF)0160
|2 StatID
|b Essential Science Indicators
|d 2025-11-07
915 _ _ |a WoS
|0 StatID:(DE-HGF)0113
|2 StatID
|b Science Citation Index Expanded
|d 2025-11-07
915 _ _ |a DBCoverage
|0 StatID:(DE-HGF)0150
|2 StatID
|b Web of Science Core Collection
|d 2025-11-07
915 _ _ |a DBCoverage
|0 StatID:(DE-HGF)1030
|2 StatID
|b Current Contents - Life Sciences
|d 2025-11-07
915 _ _ |a JCR
|0 StatID:(DE-HGF)0100
|2 StatID
|b EUR J NUCL MED MOL I : 2022
|d 2025-11-07
915 _ _ |a DBCoverage
|0 StatID:(DE-HGF)0600
|2 StatID
|b Ebsco Academic Search
|d 2025-11-07
915 _ _ |a Peer Review
|0 StatID:(DE-HGF)0030
|2 StatID
|b ASC
|d 2025-11-07
915 _ _ |a IF >= 5
|0 StatID:(DE-HGF)9905
|2 StatID
|b EUR J NUCL MED MOL I : 2022
|d 2025-11-07
920 1 _ |0 I:(DE-2719)1111015
|k Clinical Research (Munich)
|l Clinical Research (Munich)
|x 0
920 1 _ |0 I:(DE-2719)1111016
|k AG Levin
|l Clinical Neurodegeneration
|x 1
920 1 _ |0 I:(DE-2719)1110007
|k AG Haass
|l Molecular Neurodegeneration
|x 2
980 _ _ |a journal
980 _ _ |a EDITORS
980 _ _ |a VDBINPRINT
980 _ _ |a I:(DE-2719)1111015
980 _ _ |a I:(DE-2719)1111016
980 _ _ |a I:(DE-2719)1110007
980 _ _ |a UNRESTRICTED

LibraryCollectionCLSMajorCLSMinorLanguageAuthor
Marc 21