Imaging

Imaging in Alzheimer‘s

Timely diagnosis of Alzheimer’s disease to enable early intervention remains a challenge: Variations in imaging need to be reduced, image quality improved, and evaluation optimized. The integration of artificial intelligence is now significantly speeding up diagnosis, allowing treatment to begin earlier.

5min

Nadine Meru and Felix Eisenhut

Published on February 27, 2026

Alzheimer’s diagnosis is changing

As the population ages, the number of people with Alzheimer’s disease is expected to continue rising. New therapeutic approaches are promising, but timely and accurate diagnosis is critical. With the introduction of disease-modifying therapies (DMTs)¹ that offer hope to patients and their families, the demand for imaging and biomarker-based diagnostics will further increase. In healthcare systems that are already under significant strain due to workforce shortages, patients often wait months for specialist appointments. Although modern digital and AI-based systems could facilitate early detection and support, they are often still in development and require time for implementation, training, and acceptance.

A reliable Alzheimer’s diagnosis requires a multimodal approach that combines neuropsychological assessment, blood-based biomarkers, and structural and functional imaging. Structural brain MRI is an established cornerstone of dementia diagnostics. It supports differential diagnosis (e.g., exclusion of brain tumors), identification of characteristic atrophy patterns, and assessment of vascular comorbidities, white matter hyperintensities, microhemorrhages, and edema [1]. Precise MR imaging is essential to speed up the diagnostic process and enable patients to begin therapy promptly.

On T2 MR imaging, water and cerebrospinal fluid appear as bright areas due to their high signal intensity.

On T2 MR imaging, water and cerebrospinal fluid appear as bright areas. This is due to their high signal intensity.

In clinical routine, MRI failures are rarely due to technological limitations. Instead, challenges arise from variability in imaging protocols, long acquisition times, motion artifacts, inconsistent image quality, and time-consuming evaluation. Additional obstacles include subtle structural changes in early Alzheimer’s disease, limited specificity due to frequent mixed dementia pathologies, high interrater variability in visual atrophy ratings, and a lack of standardization [2].

To confirm an Alzheimer’s diagnosis, lumbar puncture is still commonly used. Although the procedure is informative, it is also invasive, uncomfortable for patients, and associated with a risk of infection. As a noninvasive alternative, positron emission tomography (PET) provides valuable diagnostic information. However, PET imaging faces significant barriers to broad implementation: limited availability, which leads to long waiting times; high costs; time-consuming examinations; and interpretation that requires specialist expertise. Together, these factors restrict the widespread use of PET imaging for early Alzheimer’s detection [3].

AI brings efficiency to Alzheimer’s diagnostics

Overcoming these obstacles is crucial to translating the benefits of advanced MRI and PET imaging into broad clinical practice.

For Alzheimer’s patients, timely access to MRI can change everything. AI can support scans that are up to 50 percent faster—so more patients get imaged sooner, helping treatment decisions and follow-up.
André Hartung, President Diagnostic Imaging, Siemens Healthineers

Modern AI-powered tools can rapidly analyze large imaging datasets, reduce interpretation time, and support confident and timely diagnosis. By minimizing variability and supporting objective quantification, AI contributes to greater diagnostic consistency across sites and readers — an essential requirement in routine clinical care.

With AI-powered imaging, it takes less than two minutes to perform a complete MRI brain scan comprising five different sequences:

Beyond diagnosis, refined AI can support clinicians in selecting treatments by identifying patients who are most likely to benefit from disease-modifying therapies. Advanced classification and quantification models help stratify patients, guide personalized treatment decisions, and ultimately contribute to optimized clinical outcomes.

The disease-modifying therapies require at least five follow-up MRI scans over the course of one year. To monitor disease progression, the brain volume on follow-up scans is compared to the baseline brain MRI scan. This increases the radiologist’s workload, and evaluation can be challenging.

Integrating AI into advanced MRI and PET imaging therefore offers significant opportunities. However, challenges remain, such as high implementation costs, the need for specialized expertise, and the standardization of imaging protocols and diagnostic platforms across institutions. In addition, despite their promise, AI and deep learning also have technical and methodological limitations. These include the high dimensionality of advanced imaging data, the limited availability of large and well-annotated training datasets, and the substantial variability both within and between patients. Addressing these challenges will be key to fully realizing the potential of AI-driven imaging in Alzheimer’s diagnostics.

How AI supports differential diagnosis

AI can speed up the diagnostic process. It provides guidance for clinicians when they discuss differential diagnosis, and it can help distinguish Alzheimer’s from other types of dementia and conditions. After 12 months on disease-modifying therapies, patients with Alzheimer’s can undergo a follow-up amyloid PET scan that allows clinicians to assess therapy effect by checking the amyloid plaque burden in the brain.

This scenario shows how quality and accuracy can be achieved in diagnostic imaging to free up more time for patient interaction:

Increasing diagnostic confidence

Digital solutions support a streamlined and standardized evaluation of amyloid PET imaging as a noninvasive method for confirming an Alzheimer’s diagnosis. Advanced postprocessing enables the reliable identification and quantification of beta-amyloid plaques and tau pathology. This facilitates consistent image interpretation and assessment of pathological burden.

A PET/MR image of a patient with Alzheimer’s disease.²

The automated calculation of a Centiloid score provides an objective, reproducible measure to support clinical decision making. This standardized quantification increases diagnostic confidence and helps clinicians identify patients who are most likely to benefit from disease-modifying therapies.

A typical amyloid PET scan of a patient with Alzheimer’s disease: The red areas are the affected regions in the brain.

A typical amyloid PET scan of a patient with Alzheimer’s disease: The red areas are the affected regions in the brain.³

How AI brings efficiency to follow-up examinations

Treatment with the new disease-modifying therapies requires at least five follow-up MRI scans over 12 months. AI can provide support for reading and reporting these examinations, thereby reducing the radiologist’s workload. One example of this is the volumetric brain analysis over time that automatically segments and evaluates more than 40 different brain structures. When the results are compared to a normative database, brain areas that deviate from those of a healthy, age-matched population can be easily highlighted, and a quantitative report is generated automatically.

New diagnostic approaches with innovations using AI

AI assists clinicians in analyzing follow-up MRI scans to identify and interpret subtle changes linked to amyloid-related imaging abnormalities (ARIAs).⁴ ARIAs are changes in the brain that may appear as side effects during disease-modifying therapies. Two main types of ARIA exist: ARIA-H, which can be microbleeds or hemosiderosis, and ARIA-E, which can be edema or effusion. Using MRI to monitor patients for these side effects is critical because ARIAs can influence whether therapy is continued or whether it needs to be paused or stopped altogether. By using AI, clinicians can make timely decisions about therapy adjustments and ensure that patients receive the safest and most effective care possible.

MR images of amyloid-related imaging abnormalities (ARIA-H and ARIA-E), which are potential adverse side effects of the new disease-modifying therapies in Alzheimer’s disease.

MR images of amyloid-related imaging abnormalities (ARIA-H and ARIA-E). These are potential adverse side effects of the new disease-modifying therapies in Alzheimer’s disease.

AI-supported MRI can identify cerebral microbleeds and thereby better support standardized and reliable monitoring. This will potentially reduce errors and eliminate the need for repeat procedures, leaving more time for what really matters — the patient.

Identifying cerebral microbleeds with the help of AI (prototype)

Identifying cerebral microbleeds with the help of AI (prototype).

Conclusion: Future-oriented and patient-centered steps

Integrating AI into Alzheimer’s diagnostics is a transformative step toward more efficient, accurate, and patient-centered care. By addressing long-standing challenges in MR acquisition, image quality, and workflow standardization, AI not only accelerates the diagnostic process but also enhances diagnostic confidence and enables individualized therapy monitoring. As AI technologies continue to evolve, they promise to further empower clinicians — by streamlining routine tasks, supporting complex decision-making, and ultimately improving outcomes for patients and their families. These innovations are shaping the future of Alzheimer’s care, paving the way for earlier detection, better therapy management, and a higher quality of life. 

Find more information

By Nadine Meru and Felix Eisenhut

Nadine Meru has a PhD in biology. She works as a digital editor at Siemens Healthineers, specializing in innovative technology, digitalization and eHealth solutions, and workforce shortages.

Felix Eisenhut, MD, worked as a radiologist in the Department of Neuroradiology at Erlangen University Hospital. Today he is a strategic marketing manager for neurology at Siemens Healthineers.

¹Disease-modifying therapies (DMTs) in Alzheimer’s disease come with risks. Please refer to the specific therapy prescribing information for side effects, contraindications, and risks.

²Acquired on Biograph mMR | Study-ID: 4aaaa0025 | Image courtesy of University Hospital Plzen, Czech Republic.
Images from Biograph mMR are provided to illustrate typical PET/MR imaging capabilities and are not specific to Biograph One. Biograph One is pending 510(k) clearance and is not yet commercially available in the USA. Its future availability cannot be guaranteed.

³Data courtesy of Wentworth-Douglass Hospital, Dover, New Hampshire, USA.

⁴This application is intended for research purposes only and must not be used for patient-individual diagnostic use. The application mentioned may not be available in all countries. Please contact your Siemens Healthineers representative for more information about this research application as well as its availability in your country.

References

[1] Furtner J, Prayer D. Neuroimaging in dementia. Wien Med Wochenschr. 2021;171(11–12):274–81.

[2] Molinder A, Ziegelitz D, Maier SE, Eckerström C. Validity and reliability of the medial temporal lobe atrophy scale in a memory clinic population. BMC Neurol. 2021;21(1):289.

[3] Shi Q, Hou J, Peng X, Xu Z, Wang Y, Cao D. Magnetic resonance imaging analysis for Alzheimer's disease diagnosis using artificial intelligence: Methods, challenges, and opportunities. Ageing Res Rev. 2026;113:102943. Epub 2025 Nov 19.

[4] U.S. Food & Drug Administration. FDA to recommend additional, earlier MRI monitoring for patients with Alzheimer’s disease taking Leqembi (lecanemab). Current as of August 28, 2025. Accessed January 1, 2026. Available from: https://www.fda.gov/drugs/drug-safety-and-availability/fda-recommend-additional-earlier-mri-monitoring-patients-alzheimers-disease-taking-leqembi-lecanemab