Brain scans can predict progression to dementia

Approximately 40% of people with mild cognitive impairment develop dementia within three years of diagnosis. This rapid progression highlights the importance of having reliable tools to identify the patients at highest risk. A recent study shows that this progression can be predicted through volumetric brain analysis using scans, a widely accessible and low-cost imaging technique.

Researchers analyzed data from 791 patients with mild cognitive impairment. Among them, 40% progressed to dementia within three years. The results reveal that individuals who developed dementia were on average older, more often women, and carriers of a genetic variant known to increase the risk of Alzheimer’s disease. Their initial cognitive state, assessed by clinical tests, was also more impaired.

Analysis of the scans revealed significant differences in certain brain regions. Patients who progressed to dementia showed greater enlargement of the inferior lateral ventricles and spaces filled with cerebrospinal fluid in the parietal and occipital areas. These changes reflect brain atrophy, i.e., a reduction in brain volume, often associated with the progression of Alzheimer’s disease.

To predict this progression, scientists developed a model based on artificial intelligence. This model combines clinical data, such as age or cognitive test results, with volumetric measurements obtained from the scans. The model achieved remarkable accuracy, correctly identifying at-risk patients in nearly 84% of cases. Six factors proved particularly decisive: age, the genetic variant, the level of initial cognitive impairment, and three specific measurements of brain atrophy in the mentioned regions.

Using scans offers a major advantage: this technique is already commonly used in clinical practice to rule out other causes of cognitive disorders, such as tumors or vascular accidents. Thus, integrating this volumetric analysis would not require additional tests, making this approach both practical and cost-effective. Unlike other methods, such as MRI or cerebrospinal fluid biomarkers, scans are accessible in most hospital centers and do not require invasive procedures.

The results show that this approach not only accurately predicts progression to dementia but also identifies the most influential factors. For example, atrophy in the inferior lateral ventricles, an area near the hippocampus, is a strong indicator of increased risk. The hippocampus plays a key role in memory, and its shrinkage is often observed in the early stages of Alzheimer’s disease. Similarly, changes in the parietal and occipital regions, linked to neural networks involved in complex cognitive functions, confirm their importance in disease progression.

This model could have a significant impact on patient care. By identifying those at high risk of developing dementia in the short term, doctors could tailor follow-up, offer early interventions, and better inform patients and their loved ones. Conversely, individuals at low risk could avoid unnecessary tests or overly frequent consultations. This risk stratification could also improve the efficiency of clinical trials by targeting participants most likely to show rapid progression, facilitating the development of new treatments.

The study confirms that the scan, often considered a basic diagnostic tool, can become a key element for prognosis. By leveraging already available data, this method offers an accessible solution for personalized medicine, without increasing the burden on patients or healthcare systems.

Media Sources

Reference Document

DOI: https://doi.org/10.1038/s41598-026-45439-8

Title: Scalable CT-based prognostic modeling of dementia conversion in mild cognitive impairment

Journal: Scientific Reports

Publisher: Springer Science and Business Media LLC

Authors: Seongbeom Park; Jehyun Ahn; Duk L. Na; Hee Jin Kim; Hyemin Jang; Jun Pyo Kim; Sung Hoon Kang; Jihwan Yun; Min Young Chun; Sang Won Seo; Kichang Kwak