Session Highlight: 2026 PAS Congress

How neuroimaging is rapidly transforming the diagnosis and treatment of Parkinson’s disease

---

Experts share how multimodal and alpha-synuclein imaging can offer deeper insights into subtle neurodegeneration across the disease spectrum.

Article Overview Jump to:  How is Artificial Intelligence Currently Used in Imaging Tools for Parkinson’s Disease? How Multimodal Imaging Aids in Early Detection and Monitoring Disease Progression in PD? Relevance of Alpha-Synuclein Imaging in Synucleinopathies References

 

How is artificial intelligence currently used in imaging tools for Parkinson’s disease?

Artificial intelligence (AI) is rapidly reshaping how we use neuroimaging to diagnose and understand Parkinson’s disease (PD).

Traditionally, clinical imaging has focused on confirming dopaminergic loss, most commonly with DAT-SPECT to assess nigrostriatal integrity. While this technique is highly accurate for detecting presynaptic dopaminergic deficits in established PD [1], conventional dopaminergic imaging with visual interpretation and semi-quantitative measures has limitations, particularly in early/prodromal disease, tracking progression, differentiating PD from atypical parkinsonian syndromes, and capturing PD’s biological heterogeneity.

PD research is now shifting from symptom-based descriptions to biologically grounded frameworks [2,3]. Advanced techniques, including structural and functional connectivity analyses; cerebral glucose metabolism imaging; and MRI markers (e.g., neuromelanin-sensitive imaging, free-water diffusion MRI, etc.), offer deeper insights into subtle neurodegeneration across the disease spectrum [4].

AI unlocks hidden imaging signals

Integrating AI with neuroimaging in PD offers many potential advantages [5-7]. Neuroimaging contains rich structural, functional, and molecular information, much of it too subtle for the human eye. AI can extract and integrate these signals at scale: 1) Supervised learning models support diagnosis and predict progression; 2) Unsupervised learning identifies biologically meaningful subtypes; 3) Deep learning automatically extracts complex spatial and textural patterns.

In dopaminergic imaging, AI models have achieved diagnostic accuracies of 90-97% in distinguishing PD from healthy controls, often outperforming traditional semi-quantitative measures. These approaches can enhance sensitivity to subtle deficits, potentially help reclassify scans without dopaminergic deficits (SWEDD) [8], and reduce interobserver variability [9,10].

Neuromelanin-sensitive MRI is another promising marker to assess dopaminergic neuronal loss in the substantia nigra, but limited by time-intensive, manual segmentation analyses. Deep learning enables a fast, automated, and reproducible detection of neuromelanin-related signal loss in the substantia nigra in PD [11].

¹⁸F-FDG PET can improve differentiation of PD from atypical parkinsonian syndromes by capturing both striatal and extra-striatal metabolic changes. AI-enhanced analyses now report sensitivities and specificities exceeding 90% for distinguishing PD from atypical parkinsonisms [12,13].

Diffusion MRI provides a promising non-invasive measure of microstructural neurodegeneration. Free-water imaging quantifies tissue and extracellular changes within the substantia nigra in PD, and across widespread gray and white matter regions in atypical parkinsonisms [14-16]. In PD, consistently elevated free water in the posterior substantia nigra is a robust diagnostic and progression marker [14,15]. Recently, machine learning applied to free-water imaging metrics has been shown to accurately distinguish PD from parkinsonian-type multiple system atrophy (MSA-P) and progressive supranuclear palsy (PSP) [17,18]. Integration into clinical imaging workflows is now underway.

Challenges and the path forward

Despite its promise, AI in PD neuroimaging faces many challenges [5,12,19,20]. Diagnostic labels can be noisy. Datasets are often small or heterogeneous, and many models lack external validation and interpretability. Longitudinal and prodromal data remain limited. Future progress will depend on large, multicenter collaborations with standardized imaging protocols; multimodal integration of imaging, clinical, genetic, and digital biomarkers; and development of more transparent, auditable AI systems suitable for safe clinical deployment.

Ultimately, AI is not designed to replace clinical expertise. Used thoughtfully, it can enhance diagnostic precision, enable earlier intervention, and move the field closer to personalized care in PD.

How multimodal imaging aids in early detection and monitoring disease progression in PD?

Biomarkers play a central role in uncovering PD pathophysiology. Defined as measurable indicators of biological processes, biomarkers can support diagnosis, prognosis, assessment of susceptibility, and evaluation of therapeutic safety. Imaging-based biomarkers offer insight into mechanisms beyond clinical symptoms.

Traditional biomarker validation models rely on clinical phenotypes — such as tremor-dominant versus PIGD subtypes — as the reference standard. However, this introduces bias, as phenotypes do not sufficiently capture PD’s biological heterogeneity. The presentation contrasts this with a biomarker-driven, phenotype-agnostic model, aligned with emerging frameworks that emphasize molecular signals over clinical labels in defining biologically coherent subgroups.

The presentation reviewed a wide spectrum of neuroimaging modalities. Dopaminergic PET and SPECT techniques — such as F‑DOPA, VMAT2, and DAT imaging — detect presynaptic nigrostriatal degeneration and support differential diagnosis among parkinsonian syndromes, including PSP and MSA. Metabolic PET imaging reveals how metabolic changes propagate across neural networks, linking metabolic patterns to motor and cognitive dysfunction.

Synaptic density imaging using SV2A PET tracers (e.g., 18F‑SynVesT‑1) provides in vivo quantification of synaptic integrity and highlights early synaptic vulnerability in degenerative parkinsonisms. MRI-based markers — including neuromelanin-sensitive imaging, diffusion MRI (e.g., free-water mapping), iron-sensitive imaging (QSM, R2*), and structural indices such as the MR Parkinsonism Index — offer powerful tools for assessing nigral degeneration, distinguishing atypical syndromes, and tracking progression.

Additional modalities — such as cardiac MIBG scintigraphy for autonomic changes and TSPO PET for neuroinflammation — extend biomarker capabilities to non-dopaminergic and peripheral systems, consistent with multisystem models of PD pathology.

PD is a multisystem disorder requiring integrated biological characterization. A multimodal imaging strategy — paired with biomarker-driven disease classification — will be essential for earlier detection, improved differential diagnosis, progression monitoring, and development of targeted therapies.

Relevance of alpha-synuclein imaging in synucleinopathies

Synucleinopathies include various diseases such as Parkinson’s disease, dementia with Lewy bodies and multiple system atrophy. Their pathological hallmark is Alpha-synuclein aggregation, and differential diagnosis is still based mainly on clinical characteristics, leading to frequent misdiagnosis. A biologically based classification has been recently proposed, where synuclein plays a central role (although not always present or necessary for the diagnosis), allowing for an earlier, more precise diagnosis, and ultimately offering the possibility of early therapeutic intervention in the future.

There have been advances in biomarker research in multiple biological samples, and during the last decade, the imaging field has been under active investigation for specific alpha-syn tracers, which could offer the possibility of alpha-syn detection in an in-vivo, non-invasive way.

The development of PET ligands targeting alpha-synuclein is challenging due to the low abundance of alpha-syn in brain tissue, its intracellular localization and the similarity to other misfolded proteins in its conformational state. However, several candidate tracers currently under investigation in preclinical and clinical studies are aiming to achieve sufficient affinity and selectivity, while minimizing off-target binding.

The presentation discussed several candidate compounds including [18F]C05-05 (1), [18F]SPAL-T-06 (2), [18F]ACI-12589 (3), [11C]MODAG-005 (4), and [18F] F0502B (5). Early studies suggest that some tracers may preferentially detect alpha-syn pathology in MSA, as compared to other synucleinopathies, but additional validation is required.

Further studies are critical for advancing precision medicine, enabling visualization of pathological processes, improving diagnostic accuracy and moving towards a patient-centered approach in order to achieve the implementation of disease-modifying therapies in synucleinopathies in the future.

 

 

References

Section: How is Artificial Intelligence Currently Used in Imaging Tools for Parkinson’s Disease?

1. Hastings A, Cullinane P, Wrigley S, et al. Neuropathologic Validation and Diagnostic Accuracy of Presynaptic Dopaminergic Imaging in the Diagnosis of Parkinsonism. Neurology. Jun 11 2024;102(11):e209453. doi:10.1212/WNL.0000000000209453
2. Hoglinger GU, Adler CH, Berg D, et al. A biological classification of Parkinson's disease: the SynNeurGe research diagnostic criteria. Lancet Neurol. Feb 2024;23(2):191-204. doi:10.1016/S1474-4422(23)00404-0
3. Simuni T, Chahine LM, Poston K, et al. A biological definition of neuronal alpha-synuclein disease: towards an integrated staging system for research. Lancet Neurol. Feb 2024;23(2):178-190. doi:10.1016/S1474-4422(23)00405-2
4. Zarkali A, Thomas GEC, Zetterberg H, Weil RS. Neuroimaging and fluid biomarkers in Parkinson's disease in an era of targeted interventions. Nat Commun. Jul 5 2024;15(1):5661. doi:10.1038/s41467-024-49949-9
5. Dennis AP, Strafella AP. The role of AI and machine learning in the diagnosis of Parkinson's disease and atypical parkinsonisms. Parkinsonism Relat Disord. Sep 2024;126:106986. doi:10.1016/j.parkreldis.2024.106986
6. Gupta R, Kumari S, Senapati A, Ambasta RK, Kumar P. New era of artificial intelligence and machine learning-based detection, diagnosis, and therapeutics in Parkinson's disease. Ageing Res Rev. Sep 2023;90:102013. doi:10.1016/j.arr.2023.102013
7. Myszczynska MA, Ojamies PN, Lacoste AMB, et al. Applications of machine learning to diagnosis and treatment of neurodegenerative diseases. Nat Rev Neurol. Aug 2020;16(8):440-456. doi:10.1038/s41582-020-0377-8
8. Choi H, Ha S, Kang H, Lee H, Lee DS, Alzheimer's Disease Neuroimaging I. Deep learning only by normal brain PET identify unheralded brain anomalies. EBioMedicine. May 2019;43:447-453. doi:10.1016/j.ebiom.2019.04.022
9. Palumbo B, Fravolini ML, Buresta T, et al. Diagnostic accuracy of Parkinson disease by support vector machine (SVM) analysis of 123I-FP-CIT brain SPECT data: implications of putaminal findings and age. Medicine (Baltimore). Dec 2014;93(27):e228. doi:10.1097/MD.0000000000000228
10. Martinez-Murcia FJ, Gorriz JM, Ramirez J, Moreno-Caballero M, Gomez-Rio M. Parametrization of textural patterns in 123I-ioflupane imaging for the automatic detection of Parkinsonism. Med Phys. Jan 2014;41(1):012502. doi:10.1118/1.4845115
11. Gaurav R, Valabregue R, Yahia-Cherif L, et al. NigraNet: An automatic framework to assess nigral neuromelanin content in early Parkinson's disease using convolutional neural network. Neuroimage Clin. 2022;36:103250. doi:10.1016/j.nicl.2022.103250
12. Zhang J. Mining imaging and clinical data with machine learning approaches for the diagnosis and early detection of Parkinson's disease. NPJ Parkinsons Dis. Jan 21 2022;8(1):13. doi:10.1038/s41531-021-00266-8
13. Meyer PT, Frings L, Rucker G, Hellwig S. (18)F-FDG PET in Parkinsonism: Differential Diagnosis and Evaluation of Cognitive Impairment. J Nucl Med. Dec 2017;58(12):1888-1898. doi:10.2967/jnumed.116.186403
14. Ofori E, Pasternak O, Planetta PJ, et al. Increased free water in the substantia nigra of Parkinson's disease: a single-site and multi-site study. Neurobiol Aging. Feb 2015;36(2):1097-104. doi:10.1016/j.neurobiolaging.2014.10.029
15. Ofori E, Pasternak O, Planetta PJ, et al. Longitudinal changes in free-water within the substantia nigra of Parkinson's disease. Brain. Aug 2015;138(Pt 8):2322-31. doi:10.1093/brain/awv136
16. Planetta PJ, Ofori E, Pasternak O, et al. Free-water imaging in Parkinson's disease and atypical parkinsonism. Brain. Feb 2016;139(Pt 2):495-508. doi:10.1093/brain/awv361
17. Archer DB, Bricker JT, Chu WT, et al. Development and Validation of the Automated Imaging Differentiation in Parkinsonism (AID-P): A Multi-Site Machine Learning Study. Lancet Digit Health. Sep 2019;1(5):e222-e231. doi:10.1016/s2589-7500(19)30105-0
18. Vaillancourt DE, Barmpoutis A, Wu SS, et al. Automated Imaging Differentiation for Parkinsonism. JAMA Neurol. May 1 2025;82(5):495-505. doi:10.1001/jamaneurol.2025.0112
19. Garcia Santa Cruz B, Husch A, Hertel F. Machine learning models for diagnosis and prognosis of Parkinson's disease using brain imaging: general overview, main challenges, and future directions. Front Aging Neurosci. 2023;15:1216163. doi:10.3389/fnagi.2023.1216163
20. Shokrpour S, MoghadamFarid A, Bazzaz Abkenar S, Haghi Kashani M, Akbari M, Sarvizadeh M. Machine learning for Parkinson's disease: a comprehensive review of datasets, algorithms, and challenges. NPJ Parkinsons Dis. Jul 1 2025;11(1):187. doi:10.1038/s41531-025-01025-9

Section: Relevance of alpha-synuclein imaging in synucleinopathies

1. Endo H, Ono M, Takado Y, Matsuoka K, Takahashi M, Tagai K, Kataoka Y, Hirata K, Takahata K, Seki C, et al. Imaging α-synuclein pathologies in animal models and patients with Parkinson’s and related diseases. Neuron. 2024;112(15):2540–2557.e8. doi: 10.1016/j.neuron.2024.05.006
2. 2. Matsuoka, K., Ono, M., Takado, Y., Hirata, K., Endo, H., Ohfusa, T., Kojima, T., Yamamoto, T., Onishi, T., Orihara, A., Tagai, K., Takahata, K., Seki, C., Shinotoh, H., Kawamura, K., Shimizu, H., Shimada, H., Kakita, A., Zhang, M.-R., Suhara, T. and Higuchi, M. (2022), High-Contrast Imaging of α-Synuclein Pathologies in Living Patients with Multiple System Atrophy. Mov Disord, 37: 2159-2161
3. 3. Smith, R., Capotosti, F., Schain, M. et al. The α-synuclein PET tracer [18F] ACI-12589 distinguishes multiple system atrophy from other neurodegenerative diseases. Nat Commun 14, 6750 (2023)
4. 4. Saw R, Buss S, Schmidt F, Ryazanov S, Leonov A, Bleher D, Gotegerd A, Kuebler L, Roben B, Schmidt F, Reimold M, Bonanno F, Ruf V, Dahl B, Sandiego C, Henry K, Papadopoulos I, Schaller M, Kahle P, Levin J, Gasser T, Brockmann K, Reischl G, Fougere C, Pichler B, Maurer A, Griesinger G, Giese A, Herfert K. REsearch Square, 2024 (preprint)
5. 5. Xiang J, Tao Y, Xia Y, Luo S, Zhao Q, Li B, Zhang X, Sun Y, Xia W, Zhang M, Kang SS, Ahn EH, Liu X, Xie F, Guan Y, Yang JJ, Bu L, Wu S, Wang X, Cao X, Liu C, Zhang Z, Li D, Ye K. Development of an α-synuclein positron emission tomography tracer for imaging synucleinopathies. Cell. 2023 Aug 3;186(16):3350-3367.e19