Manifold diffusion transformer enables diagnostic multi-tracer PET synthesis from MRI for early detection of Alzheimer’s disease

Chenhui Wang; Sirong Piao; Jie Wang; Zhaoyang Li; Tao Chen; Mei Cui; Jun Zhao; Qihao Guo; Junping Zhang; Fang Xie; Yuxin Li; Hongming Shan

doi:10.59717/j.xinn-inform.2026.100048

The Innovation Informatics > 2026 Vol. 2 > No. 2 > 100048

Next Article Previous Article

ARTICLE Open Access

Cite PDF

Manifold diffusion transformer enables diagnostic multi-tracer PET synthesis from MRI for early detection of Alzheimer’s disease

1.
Institute of Science and Technology for Brain-inspired Intelligence, Fudan University, Shanghai 200433, China
2.
Department of Radiology, Huashan Hospital, Fudan University, Shanghai 200040, China
3.
Department of Nuclear Medicine and PET Center, Huashan Hospital, Fudan University, Shanghai 200040, China
4.
Department of Neurology, Huashan Hospital, Fudan University, Shanghai 200040, China
5.
Department of Nuclear Medicine, Shanghai East Hospital, Tongji University School of Medicine, Shanghai 200123, China
6.
Department of Gerontology, Shanghai Jiao Tong University Affiliated Sixth People's Hospital, Shanghai 200233, China
7.
Shanghai Key Lab of Intelligent Information Processing, Fudan University, Shanghai 200433, China
8.
College of Computer Science and Artificial Intelligence, Fudan University, Shanghai 200433, China
9.
Institute of Functional and Molecular Medical Imaging, Fudan University, Shanghai 200040, China
10.
MOE Frontiers Center for Brain Science, Fudan University, Shanghai 200433, China
11.
Key Laboratory of Computational Neuroscience and Brain-Inspired Intelligence (Ministry of Education), Fudan University, Shanghai, 200433, China
12.
These authors contributed equally

Show all affliationsShow less

More Information

Corresponding authors: fangxie@fudan.edu.cn (F.X.); liyuxin@fudan.edu.cn (Y.L.); hmshan@fudan.edu.cn (H.S.)
Received Date: March 03, 2026

Accepted Date: May 09, 2026

Published Online: May 14, 2026

The Innovation Informatics 2(2), https://doi.org/10.59717/j.xinn-inform.2026.100048 | Cite this article

GRAPHICAL ABSTRACT

GRAPHICAL ABSTRACT

DownLoad: Full size image

PUBLIC SUMMARY

PUBLIC SUMMARY
1. AI transforms magnetic resonance imaging (MRI) into multi-tracer positron emission tomography (PET).
  
  It visualizes key Alzheimer’s disease biomarkers to characterize the full neurodegenerative spectrum.
  
  These synthetic PET images show high structural fidelity and strong clinical agreement with real scans.
  
  As a cost-effective screening tool, synthetic multi-tracer PET greatly enhances standalone MRI diagnostics.

ABSTRACT

ABSTRACT

Positron emission tomography (PET) plays a critical role in the early detection of Alzheimer's disease (AD) by revealing pathophysiological changes prior to clinical onset or neurodegeneration detectable by magnetic resonance imaging (MRI). However, the clinical deployment of multi-tracer PET is limited by high cost and restricted accessibility, especially in resource-limited settings. Here, we introduce MaM-DiT, a novel modality-aware manifold diffusion transformer that synthesizes high-quality, diagnostically informative multi-tracer PET images directly from MRI. By capturing spatiotemporal features across the amyloid–tau–neurodegeneration (A−T−N) spectrum, MaM-DiT enables efficient, non-invasive assessment of AD pathology. The model was trained on 869 paired MRI–PET scans from three clinical centers, visually validated on 106 samples from an independent center, and diagnostically evaluated on 1,207 subjects from the ADNI cohort. We comprehensively assessed the image quality and diagnostic utility of synthetic PETs and found that they exhibited high visual fidelity to real scans across tracers (SSIM > 0.87). Moreover, synthetic PETs enabled AD subtype stratification and improved MRI-along diagnostic performance, yielding gains of up to 13.8% in AUC-ROC and 12.5% in AUC-PR. These results highlight the potential of AI-synthesized multi-tracer PET as a cost-effective prescreening tool to triage at-risk individuals, supporting early detection and optimized resource allocation in diverse healthcare settings.
- Positron Emission Tomography (PET) /
- Magnetic Resonance Imaging (MRI) /
- Multi-center Study /
- Alzheimer's disease (AD) /
- Diffusion Transformer

FullText(HTML)

REFERENCES

[1]	Organization W.H. Fact sheets of dementia. Geneva, Switzerland, 2021. https://www.who.int/news-room/fact-sheets/detail/dementia View in Article Google Scholar
[2]	Parra M.A., Baez S., Allegri R., et al. (2018). Dementia in latin america: Assessing the present and envisioning the future. Neurology 90:222−231. DOI:10.1212/WNL.0000000000004897 View in Article CrossRef Google Scholar
[3]	Ferri C.P., Prince M., Brayne C., et al. (2005). Global prevalence of dementia: A Delphi consensus study. Lancet 366:2112−2117. DOI:10.1016/S0140-6736(05)67889-0 View in Article CrossRef Google Scholar
[4]	Chouliaras L. and O’Brien J.T. (2023). The use of neuroimaging techniques in the early and differential diagnosis of dementia. Mol. Psychiatry 28:4084−4097. DOI:10.1038/s41380-023-02215-8 View in Article CrossRef Google Scholar
[5]	Frisoni G.B., Hansson O., Nichols E., et al. (2025). New landscape of the diagnosis of Alzheimer’s disease. Lancet 406:1389−1407. DOI:10.1016/S0140-6736(25)01294-2 View in Article CrossRef Google Scholar
[6]	Schmechel D., Saunders A., Strittmatter W., et al. (1993). Increased amyloid beta-peptide deposition in cerebral cortex as a consequence of apolipoprotein E genotype in late-onset Alzheimer disease. Proc. Natl. Acad. Sci. USA 90:9649−9653. DOI:10.1073/pnas.90.20.9649 View in Article CrossRef Google Scholar
[7]	Jack Jr C.R., Bennett D.A., Blennow K., et al. (2018). NIA-AA research framework: Toward a biological definition of Alzheimer’s disease. Alzheimers. Dement. 14:535−562. DOI:10.1016/j.jalz.2018.02.018 View in Article CrossRef Google Scholar
[8]	Hanseeuw B.J., Betensky R.A., Jacobs H.I., et al. (2019). Association of amyloid and tau with cognition in preclinical Alzheimer disease: A longitudinal study. JAMA Neurol. 76:915−924. DOI:10.1001/jamaneurol.2019.1424 View in Article CrossRef Google Scholar
[9]	Maruyama M., Shimada H., Suhara T., et al. (2013). Imaging of tau pathology in a tauopathy mouse model and in Alzheimer patients compared to normal controls. Neuron 79:1094−1108. DOI:10.1016/j.neuron.2013.07.037 View in Article CrossRef Google Scholar
[10]	Lowe V.J., Lundt E.S., Albertson S.M., et al. (2020). Tau-positron emission tomography correlates with neuropathology findings. Alzheimers. Dement. 16:561−571. DOI:10.1016/j.jalz.2019.09.079 View in Article CrossRef Google Scholar
[11]	Ryan N.S., Rossor M.N. and Fox N.C. (2015). Alzheimer’s disease in the 100 years since Alzheimer’s death. Brain 138:3816−3821. DOI:10.1093/brain/awv316 View in Article CrossRef Google Scholar
[12]	DeTure M.A. and Dickson D.W. (2019). The neuropathological diagnosis of Alzheimer’s disease. Mol. Neurodegener. 14:32. DOI:10.1186/s13024-019-0333-5 View in Article CrossRef Google Scholar
[13]	Guo T., Korman D., Baker S.L., et al. (2021). Longitudinal cognitive and biomarker measurements support a unidirectional pathway in Alzheimer’s disease pathophysiology. Biol. Psychiatry 89:786−794. DOI:10.1016/j.biopsych.2020.06.029 View in Article CrossRef Google Scholar
[14]	Mill J. and Li L. (2022). Recent advances in understanding of Alzheimer’s disease progression through mass spectrometry-based metabolomics. Phenomics 2:1−17. DOI:10.1007/s43657-021-00036-9 View in Article CrossRef Google Scholar
[15]	Wang C., Lei Y., Chen T., et al. (2024). HOPE: Hybrid-granularity ordinal prototype learning for progression prediction of mild cognitive impairment. IEEE J. Biomed. Health Inform. 28: 6429-6440. DOI: 10.1109/JBHI.2024.3357453 DOI:10.1109/JBHI.2024.3357453 View in Article CrossRef Google Scholar
[16]	Zhang Z.-C., Zhao X., Dong G., et al. (2023). Improving Alzheimer’s disease diagnosis with multi-modal PET embedding features by a 3D multi-task MLP-mixer neural network. IEEE J. Biomed. Health Inform. 27:4040−4051. DOI:10.1109/JBHI.2023.3280823 View in Article CrossRef Google Scholar
[17]	Wang C., Piao S., Chen Z., et al. (2026). E2AD: Enhanced and explainable Alzheimer’s disease detection framework via anatomy-and relation-aware cross-modal knowledge distillation. Med. Image Anal. 104099. DOI:10.1016/j.media.2026.104099 View in Article CrossRef Google Scholar
[18]	Wang C., Piao S., Huang Z., et al. (2024). Joint learning framework of cross-modal synthesis and diagnosis for Alzheimer’s disease by mining underlying shared modality information. Med. Image Anal. 91:103032. DOI:10.1016/j.media.2023.103032 View in Article CrossRef Google Scholar
[19]	Ou Z., Jiang C., Pan Y., et al. (2024) A prior-information-guided residual diffusion model for multimodal PET synthesis from MRI. IJCAI. 2024: 4769–4777. DOI: 10.24963/ijcai.2024/527 View in Article Google Scholar
[20]	Lee J., Burkett B.J., Min H.-K., et al. (2024). Synthesizing images of tau pathology from crossmodal neuroimaging using deep learning. Brain 147:980−995. DOI:10.1093/brain/awad346 View in Article CrossRef Google Scholar
[21]	Ou Z., Pan Y., Li Y., et al. (2024) Synthesizing Aβ-PET via an image and label conditioning latent diffusion model for detecting amyloid status. ICASSP. 2024: 6610–6614. DOI:10.1109/ICASSP48485.10448346 View in Article CrossRef Google Scholar
[22]	Goodfellow I., Pouget-Abadie J., Mirza M., et al.(2014). Generative adversarial nets. NeurIPS. DOI: 10.48550/arXiv.1406.2661 View in Article Google Scholar
[23]	Ho J., Jain A. and Abbeel P. (2020). Denoising diffusion probabilistic models. NeurIPS. 2020: 6840–6851. DOI: 10.48550/arXiv.2006.11239 View in Article Google Scholar
[24]	Lundeen T.F., Seibyl J.P., Covington M.F., et al. (2018). Signs and artifacts in amyloid PET. Radiographics 38:2123−2133. DOI:10.1148/rg.2018180160 View in Article CrossRef Google Scholar
[25]	Rombach R., Blattmann A., Lorenz D., et al. (2022). High-resolution image synthesis with latent diffusion models. CVPR. 2022: 10684–10695. DOI: 10.1109/CVPR52688.2022.01042 View in Article Google Scholar
[26]	Ma C., Ji Y., Ye J., et al. (2025). MedITok: A unified tokenizer for medical image synthesis and interpretation. arXiv preprint. DOI: 10.48550/arXiv.2505.19225 View in Article Google Scholar
[27]	Johnson J., Alahi A. and Li F. (2016). Perceptual losses for real-time style transfer and superresolution. in: ECCV. 2016: 694–711. DOI: 10.1007/978-3-319-46475-6_43 View in Article Google Scholar
[28]	Kingma D.P. and Welling M. (2014). Auto-encoding variational bayes. ICLR. DOI: 10.48550/arXiv.1312.6114 View in Article Google Scholar
[29]	Mao X., Li Q., Xie H., et al. (2017) Least squares generative adversarial networks. in: ICCV. 2017: 2794–2802. DOI: 10.1109/ICCV.2017.304 View in Article Google Scholar
[30]	Lin Y., Wang C., Peng Z., et al. (2026). UniField: A unified field-aware MRI enhancement framework. arXiv:2603.09223. DOI: 10.48550/arXiv.2603.09223 View in Article Google Scholar
[31]	Wang C., Zheng B., Bao L., et al. (2026). Brain-WM: Brain glioblastoma world model. arXiv:preprint. DOI: 10.48550/arXiv.2603.07562 View in Article Google Scholar
[32]	Huang X. and Belongie S. (2017). Arbitrary style transfer in real-time with adaptive instance normalization. ICCV. 2017:1501–1510. DOI:10.1109/ICCV.2017.167 View in Article Google Scholar
[33]	Cao B., Yao X., Wang C., et al. (2026). Dynamic differential linear attention: Enhancing linear diffusion transformer for high-quality image generation. arXiv: preprint. DOI: 10.48550/arXiv.2601.13683 View in Article Google Scholar
[34]	Dao T., Fu D., Ermon S., et al. (2022). Flash-attention: Fast and memory-efficient exact attention with IO-awareness. NeurIPS 35:16344−16359. DOI:10.48550/arXiv.2205.14135 View in Article CrossRef Google Scholar
[35]	Zhao X., Zhao Z., Liu Z., et al. (2024). OpenDiT: An easy, fast and memory-efficient system for DiT training and inference. https://github.com/NUS-HPC-AI-Lab/OpenDiT.2024 View in Article Google Scholar
[36]	Song J., Meng C. and Ermon S. (2021). Denoising Diffusion Implicit Models. ICLR. 2021. DOI: 10.48550/arXiv.2010.02502 View in Article Google Scholar
[37]	Qiu S., Miller M.I., Joshi P.S., et al. (2022). Multimodal deep learning for Alzheimer’s disease dementia assessment. Nat. Commun. 13:3404. DOI:10.1038/s41467-022-31037-5 View in Article CrossRef Google Scholar
[38]	Fischl B. (2012). FreeSurfer. NeuroImage 62:774−781. DOI:10.1016/j.neuroimage.2012.01.021 View in Article CrossRef Google Scholar
[39]	Breiman L. (2001). Random forests. Mach. Learn. 45:5−32. DOI:10.1023/A:1010933404324 View in Article CrossRef Google Scholar
[40]	Jack C.R., Knopman D.S., Jagust W.J., et al. (2013). Tracking pathophysiological processes in Alzheimer’s disease: An updated hypothetical model of dynamic biomarkers. Lancet Neurol. 12:207−216. DOI:10.1016/S1474-4422(12)70291-0 View in Article CrossRef Google Scholar
[41]	Donohue M.C., Sperling R.A., Petersen R., et al. (2017). Association between elevated brain amyloid and subsequent cognitive decline among cognitively normal persons. JAMA 317:2305−2316. DOI:10.1001/jama.2017.6669 View in Article CrossRef Google Scholar
[42]	Grajauskas L.A., Siu W., Medvedev G., et al. (2019). MRI-based evaluation of structural degeneration in the ageing brain: Pathophysiology and assessment. Ageing Res. Rev. 49:67−82. DOI:10.1016/j.arr.2018.11.004 View in Article CrossRef Google Scholar
[43]	Pan Y., Liu M., Xia Y., et al. (2021). Disease-image-specific learning for diagnosis-oriented neuroimage synthesis with incomplete multi-modality data. IEEE Trans. Pattern Anal. Mach. Intell. 44:6839−6853. DOI:10.1109/TPAMI.2021.3091214 View in Article CrossRef Google Scholar
[44]	Shen L., Zhao W. and Xing L. (2019). Patient-specific reconstruction of volumetric computed tomography images from a single projection view via deep learning. Nat. Biomed. Eng. 3:880−888. DOI:10.1038/s41551-019-0466-4 View in Article CrossRef Google Scholar
[45]	Salehjahromi M., Karpinets T.V., Sujit S.J., et al. (2024). Synthetic PET from CT improves diagnosis and prognosis for lung cancer: Proof of concept. Cell Rep. Med. 5:101463. DOI:10.1016/j.xcrm.2024.101463 View in Article CrossRef Google Scholar
[46]	Pinaya W.H., Graham M.S., Kerfoot E., et al. (2023). Generative AI for medical imaging: Extending the MONAI framework. arXiv: preprint. DOI: 10.48550/arXiv.2307.15208 View in Article Google Scholar
[47]	McHugh M.L. (2012). Interrater reliability: The kappa statistic. Biochem. Medica. 22:276−282. DOI:10.11613/BM.2012.031 View in Article CrossRef Google Scholar
[48]	Ronneberger O., Fischer P. and Brox T. (2015). U-Net: Convolutional networks for biomedical image segmentation. in: MICCAI. 2015: 234–241. DOI:10.1007/978-3-319-24574-4_28 View in Article Google Scholar
[49]	Hatamizadeh A., Nath V., Tang Y., et al. (2021). Swin UNETR: Swin transformers for semantic segmentation of brain tumors in MRI images. in: MICCAI workshop. 2021:272-284. DOI: 10.1007/978-3-031-08999-2_22 View in Article Google Scholar
[50]	Yu B., Zhou L., Wang L., et al. (2019). Ea-GANs: Edge-aware generative adversarial networks for cross-modality MR image synthesis. IEEE Trans. Med. Imaging 38:1750−1762. DOI:10.1109/TMI.2019.2895894 View in Article CrossRef Google Scholar
[51]	Ou Z., Pan Y., Xie F., et al. (2025). Image-and-label conditioning latent diffusion model: synthesizing Aβ-PET from MRI for detecting amyloid status. IEEE J. Biomed. Health Inform. 29:1221−1231. DOI:10.1109/JBHI.2024.3492020 View in Article CrossRef Google Scholar
[52]	Kim J. and Park H. (2024). Adaptive latent diffusion model for 3D medical image to image translation: Multi-modal magnetic resonance imaging study. in: WACV. 2024:7604–7613. DOI: 10.1109/WACV57701.2024.00743 View in Article Google Scholar
[53]	Pagani M., Nobili F., Morbelli S., et al. (2017). Early identification of MCI converting to AD: A FDG PET study. Eur. J. Nucl. Med. Mol. Imaging 44:2042−2052. DOI:10.1007/s00259-017-3761-x View in Article CrossRef Google Scholar

Cited by

ABOUT THIS ARTICLE

Cite this article:

Wang C., Piao S., Wang J., et al. (2026). Manifold diffusion transformer enables diagnostic multi-tracer PET synthesis from MRI for early detection of Alzheimer’s disease. The Innovation Informatics 2:100048. https://doi.org/10.59717/j.xinn-inform.2026.100048

Wang C., Piao S., Wang J., et al. (2026). Manifold diffusion transformer enables diagnostic multi-tracer PET synthesis from MRI for early detection of Alzheimer’s disease. The Innovation Informatics 2:100048. https://doi.org/10.59717/j.xinn-inform.2026.100048

Welcome!

To request copyright permission to republish or share portions of our works, please visit Copyright Clearance Center's (CCC) Marketplace website at marketplace.copyright.com.

PDF

Figures(7)

Supplementary Information

download_supplementary
SI-Extended-TII-DOI-48

Article Metrics

Article views(782) PDF downloads(197)

Manifold diffusion transformer enables diagnostic multi-tracer PET synthesis from MRI for early detection of Alzheimer’s disease

GRAPHICAL ABSTRACT

PUBLIC SUMMARY

ABSTRACT