AI-powered Surreal-GAN identifies diverse brain aging patterns

From a recent study published in Naturopathyresearchers are using artificial intelligence (AI)-powered ‘Surreal-GAN’, a deep representation learning model, to explore the heterogeneity of aging brains.

Study: Brain aging patterns in a large and diverse cohort of 49,482 individuals. Image credits: Wirestock Creators / Shutterstock.com

The neurology of aging

The aging human brain undergoes many structural changes that vary based on genetics, an individual’s lifestyle, and the presence of coexisting diseases. This heterogeneity in brain aging is also influenced by factors that can exacerbate or protect age-related neuropathological processes.

Early and subtle changes in specific brain regions can occur during the preclinical phases of neurodegenerative diseases such as Alzheimer’s disease. Therefore, it is critical to understand these neuroanatomical changes across a broad spectrum of individuals.

Traditional neuroimaging studies have provided important insights into the role of aging and disease in altering brain structure and function, often based on case-control comparisons. However, these methods are limited in their ability to address individual heterogeneity because they typically focus on average patterns, rather than capturing the diverse neuroanatomical changes between individuals.

Machine learning (ML) approaches have been successfully used to identify neuroimaging biomarkers of brain aging at the individual level. However, these methods often do not take into account the underlying heterogeneity and, as a result, only identify biomarkers that reflect a typical or average pattern.

About the study

The current study used Surreal-GAN to analyze patterns of brain aging by learning one-to-many transformations from a reference population (REF) to a target population (TAR) to capture heterogeneous brain changes. These data are then distilled into representational indices (R) that quantify the severity of individualized brain changes along different dimensions.

The REF group included 1,150 participants between 20 and 49 years old, while the TAR group included 8,992 individuals between 50 and 97 years old, including those with mild cognitive impairment (MCI) or dementia.

To ensure robustness, the model’s reproducibility was tested on an independent training set and between genders. The researchers also associated R indices with demographic, clinical, neurocognitive, lifestyle and genetic measures using partial correlations, voxel-based morphometry and genome-wide association studies (GWAS).

Associations with chronic diseases were examined using multiple linear regression analyzes adjusting for both age and sex. Prognostic potential was assessed using Cox proportional hazards models based on longitudinal data.

Findings of the study

Five different R indices of R1, R2, R3, R4 and R5 represented subcortical, medial temporal lobe, parieto-temporal, diffuse cortical and perisylvian atrophy, respectively. These indices were significantly associated with demographic variables, such as age and gender, as well as with several chronic diseases, including MCI, dementia, schizophrenia and Parkinson’s disease.

R2, R3, and R5 were strongly correlated with Alzheimer’s disease biomarkers, specifically cerebrospinal fluid (CSF)-pTau181 and CSF-Aβ42. R5 has been linked to a wide range of chronic diseases and lifestyle factors, such as alcohol consumption and smoking.

Baseline R indices were strong predictors of disease progression from cognitively normal to MCI and from MCI to dementia, as well as mortality risk.

Several genetic loci were associated with these R indices, including some that had not previously been linked to clinical features. Thus, R-indices can both capture the heterogeneity of brain aging and be valuable markers for understanding the progression of neurodegenerative diseases, as well as the impact of lifestyle and genetic factors on brain health.

The current study is enhanced by the use of an improved Surreal-GAN methodology with broad applicability for discovering patterns of brain aging, providing an extensible system applicable to diverse research questions. However, notable limitations of the current study include the underrepresentation of uncommon pathologies and certain diseases, as well as the chosen age limit, which may obscure some preclinical changes in the REF group.

Conclusions

The researchers developed Surreal-GAN, a five-dimensional representation system to characterize the neuroanatomical heterogeneity of brain aging. Surreal-GAN provides a novel tool for parsing the heterogeneity of brain atrophy and understanding its associations with demographic, pathological, lifestyle factors, and genetic variants. Moreover, Surreal-GAN also has the potential to improve personalized diagnostics and patient management and improve the precision and effectiveness of clinical trials.