% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@ARTICLE{Li:285004,
author = {Li, Chenyang and Jiao, Fangyang and Wu, Shaoyou and Wang,
Chenhan and Wei, Min and Zhang, Shuoyan and Wang, Luyao and
Huang, Yu and Yin, Yafu and Tian, Rong and Bernhardt,
Alexander and Katzdobler, Sabrina and Levin, Johannes and
Höglinger, Günter U and Brendel, Matthias and Rominger,
Axel and Shi, Kuangyu and Zuo, Chuantao and Jiang, Jiehui},
title = {{E}nhancing interpretability of {AI} with radiomics-based
deep neural network: proof of concept in the classification
of {P}arkinsonian syndromes with 18{F}-{FDG} {PET} imaging.},
journal = {European journal of nuclear medicine and molecular imaging},
volume = {53},
number = {3},
issn = {1619-7070},
address = {Heidelberg [u.a.]},
publisher = {Springer-Verl.},
reportid = {DZNE-2026-00138},
pages = {1962 - 1979},
year = {2026},
abstract = {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.},
keywords = {Humans / Fluorodeoxyglucose F18 / Parkinsonian Disorders:
diagnostic imaging / Parkinsonian Disorders: classification
/ Positron-Emission Tomography / Male / Female / Neural
Networks, Computer / Middle Aged / Aged / Proof of Concept
Study / Deep Learning / Image Processing, Computer-Assisted:
methods / Radiomics / 18F-fluorodeoxyglucosePET imaging
(Other) / Dual-channel neural network (Other) / Model
interpretability (Other) / Parkinsonian syndromes (Other) /
Radiomics (Other) / SHAP (Other) / Fluorodeoxyglucose F18
(NLM Chemicals)},
cin = {Clinical Research (Munich) / AG Levin / AG Haass},
ddc = {610},
cid = {I:(DE-2719)1111015 / I:(DE-2719)1111016 /
I:(DE-2719)1110007},
pnm = {353 - Clinical and Health Care Research (POF4-353) / 352 -
Disease Mechanisms (POF4-352)},
pid = {G:(DE-HGF)POF4-353 / G:(DE-HGF)POF4-352},
typ = {PUB:(DE-HGF)16},
pubmed = {pmid:40932612},
doi = {10.1007/s00259-025-07478-7},
url = {https://pub.dzne.de/record/285004},
}