ADRD

Alzheimer’s Disease and Related Dementia (ADRD)

We are among the first groups showing importance of alteration of the resting-state fMRI brain network in Alzheimer’s disease (AD) (Khazaee et al., 2015). We demonstrated alteration of the brain network in early stages of AD and mild cognitive impairment (MCI) (Hojjati et al., 2019). We demonstrated that integrating multimodal neuroimaging data (MRI, fMRI, and PET) with AI can predict neuropsychological scores in AD, suggesting a robust framework for early disease detection (Hojjati et al., 2021).We further identified key biomarkers and brain regions associated with cognitive decline in AD, providing insights into disease progression and potential therapeutic targets (Hojjati et al., 2024).

In an ongoing project, we are developing, training, and testing advanced 3D CNN-based deep learning models to predict brain age from MRI scans in a large cohort of cognitively normal (CN) subjects. By comparing the difference between chronological and predicted brain age—referred to as the “age gap”—we aim to demonstrate that this gap is significantly larger in subjects with neurodegenerative diseases such as AD, LBD, and Parkinson’s disease (PD). This suggests accelerated brain aging associated with these conditions, enhancing our understanding of age-related neuroanatomical changes in neurodegenerative diseases.

In an ongoing project, we are using deep learning–based brain-age modeling to characterize neurodegeneration in Alzheimer’s disease (AD) with Lewy body (LB) co-pathology. We develop and validate 3D CNN models on large, well-characterized neuroimaging cohorts to learn patterns of normative brain aging from structural MRI. We then apply these models to biomarker-defined clinical cohorts to quantify deviations from typical aging (i.e., brain age “gap”) across AD and LB pathophysiological profiles. Using interpretable AI methods and complementary region-of-interest analyses, we aim to identify neuroanatomical signatures associated with co-pathology and determine how these imaging-derived measures relate to longitudinal cognitive trajectories. Ultimately, this work seeks to refine biomarker-informed stratification and improve our understanding of heterogeneity in AD-related neurodegeneration.

Related Publications