Search this site
Embedded Files
NIDLL - NeuroImaging with Deep Learning Lab
  • Home
  • About
  • Members
  • Pictures/Activities
  • Research
  • Publications
  • News
  • Join Us
  • Collaborators
NIDLL - NeuroImaging with Deep Learning Lab

Research


NeuroImaging with Deep Learning Lab (NIDLL )

Neuroelectrophysiological aging clock (EEG) associated with sleep disorders, comorbidity and brain structural changes

We have developed a novel sleep EEG-based brain age index (BAI) using artificial intelligence to objectively measure brain aging influenced by sleep patterns. Unlike traditional models that rely on structural MRI data, our approach captures dynamic brain functions during sleep, providing unique insights into the aging process. Our brain age prediction model achieves a mean absolute error of 4.8 years, outperforming existing EEG-based biomarkers.

Our research highlights the significant role of sleep architecture—particularly the depth and continuity of sleep—in brain aging. We found that chronic sleep disorders like obstructive sleep apnea (OSA) and insomnia accelerate brain aging, emphasizing the critical importance of diagnosing and treating these conditions to preserve brain health.

Additionally, we discovered that shift workers exhibit specific delta wave patterns linked to accelerated brain aging, offering the first empirical evidence of the long-term brain health risks associated with shift work. We also explored the therapeutic potential of continuous positive airway pressure (CPAP) therapy for individuals with OSA. Our findings show that CPAP treatment can effectively halt the progression of brain aging, underscoring its potential to protect against the long-term consequences of sleep-disordered breathing on brain health.

Imaging of Glymphatic Function in Human Brains

The glymphatic system is a brain fluid-clearance pathway that facilitates the removal of metabolic waste by cerebrospinal fluid (CSF) through periarterial spaces (PAS) and perivenous spaces (PVS). While existing non-invasive imaging techniques like DTI-ALPS index provide whole-brain measures of glymphatic function, they lack region-specific insights critical for understanding neurodegenerative diseases. We propose a novel metric, RaPPiD (Ratio of Perpendicular to Parallel Periarterial Diffusivity), which uses diffusion MRI and T2-weighted imaging to quantify regional glymphatic function by analyzing water diffusivity at voxels of PAS. RaPPiD evaluates the ratio of perpendicular-to-parallel diffusivity, reflecting fluid movement into interstitial spaces.

In a cohort of 579 subjects from the HCP-Aging study, we found significant associations between RaPPiD values and sleep quality indices in the frontoparietal and limbic networks, as well as cardiovascular factors like systolic blood pressure in the dorsal attention network.

RaPPiD provides a refined, region-specific measure of glymphatic function, highlighting its sensitivity to sleep and vascular health. This approach may improve understanding of age-related neurodegeneration and brain clearance mechanisms. 

Motion Artifact Correction using Deep Learning

Head motion during MRI acquisition presents significant problems for subsequent neuroimaging analyses. In our most up-to-date work (Sun et al., ISMRM 2025), we aim to develop a Motion-Adaptive Diffusion Model (MADM) to correct motion artifacts in MR, improving image quality.  MADM is based on a diffusion model. Gaussian noise is added in the forward process, and a U-Net progressively denoises the images in reverse process. Our model was trained on the MR-ART dataset.


Diffusion Model

What we can do: A generative model used to   create images by simulation  the diffusion process. Key Idea: Start with random noise, and gradually transform it into a meaningful image. Forward Process: The model adds noise to the images over multiple steps, simulating a diffusion process. Reverse Process: The model learns to reverse the diffusion process, starting from noise and progressively generating a structured output via neural networks

Brain Age Prediction in Various Neurodegenerative Diseases

The focuse of this study is the development of advanced AI methods to predict brain ages in various neurological conditions and understand whether brain aging can be accelerated by those clinical conditions and reversed by treatment or rehabilitation and whether regional pattern of brain age can  be modified via brain re-organization and disease compensation mechnisms. Using idividual T1-weighted MRI scans, we extract brain (cortical) surface models, define regions of interest (ROIs), and calculate the regional brain age for each ROI. This method provides a detailed understanding of the pattern of specific brain regions that are affected by various disease progression. Currently, we are applying this technique to study healthy older adults with APOE 4 gene (Im et al., Brain Communications 2024) or Cardiometabolic syndromes (Kang et al., Alzheimer's Research & Therapy) , stroke (Park et al., medArxiv 2024), sleep disorders (including insomnia, obstructive sleep apnea [OSA], and REM sleep behavior disorder [RBD]), and mesial temporal lobe epilepsy (MTLE). For example, in a recent stroke study, we observed that the contralesional hemisphere in patients with severe motor impairment exhibited a younger brain age, suggesting a potential compensatory mechanism. These findings underscore the value of regional brain age analysis in uncovering hidden brain dynamics and guiding targeted therapeutic interventions.


Cutting-Edge Machine Learning Techniques to Analyze Disease Progression 

The focus of this study is the application of cutting-edge machine learning techniques to analyze disease progression, particularly in neurodegenerative disorders. Specifically, we utilize the Subtype and Stage Inference (SuStaIn) algorithm, an event-based probabilistic model, to classify patient populations into distinct subtypes and uncover unique progression patterns. In a recent study on Parkinson's disease (PD), we applied SuStaIn to analyze brain MRI data, focusing on cortical thickness and deep gray matter volume. This analysis revealed two distinct subtypes: a "cortex-first" subtype, characterized by cortical thinning preceding subcortical atrophy, and a "deep gray-first" subtype, where subcortical atrophy occurs earlier. These subtypes demonstrated significant differences in clinical features and disease progression rates. This research highlights the potential of data-driven methods to advance personalized medicine and enable more precise, targeted interventions.

White Matter Hyperintensities Segmentation

White matter hyperintensities (WMHs) are abnormal signals within the white matter region on the human brain MRI and have been associated with aging processes, cognitive decline, and dementia. In the current study, we proposed a U-Net with multi-scale highlighting foregrounds (HF) for WMHs segmentation. Our method, U-Net with HF, is designed to improve the detection of the WMH voxels with partial volume effects. We evaluated the segmentation performance of the proposed approach using the Challenge training dataset. Up to date, the proposed method has achieved the best overall evaluation scores, the highest dice similarity index, and the best F1-score among 39 methods submitted on the WMH Segmentation Challenge that was initially hosted by MICCAI 2017 and is continuously accepting new challengers. The evaluation of the clinical utility showed that the WMH volume that was automatically computed using U-Net with HF was significantly associated with cognitive performance and improves the classification between cognitive normal and Alzheimer's disease subjects and between patients with mild cognitive impairment and those with Alzheimer's disease. The implementation of our proposed method is publicly available using Dockerhub (https://hub.docker.com/r/wmhchallenge/pgs).

Brain Surface and Morphology Synthesis:

Changes in brain morphology, such as cortical thinning are of great value for understanding the trajectory of brain aging and various neurodegenerative diseases. In this work, we employ a generative neural network variational autoencoder (VAE) that is conditional on age and is able to generate cortical thickness maps at various ages given an input cortical thickness map. To take into account the mesh topology in the model, we propose a loss function based on weighted adjacency to integrate the surface topography defined as edge connections with the cortical thickness mapped as vertices. Compared to traditional conditional VAE that did not use the surface topological information, our method better predict “future” cortical thickness maps, especially when the age gap become wider. Our model has the potential to predict the distinctive temporo-spatial pattern of individual cortical morphology in relation to aging and neurodegenerative diseases. (ISBI 2021)

Early Brain Development

This research aims to use innovate MR Imaging techniques to quantitatively characterize normal fetal / neonatal brain development as well as atypical development associated with prematurity. This will be accomplished using a large database of MR scans which have been obtained twice per baby, as well as using various deep learning methods and imaging processing techniques, which correct head motion artifacts, accurately segment brain structures, and quantify tissue characteristics and brain structural and functional network properties. Next, development of an online platform integrated with a big data-driven deep learning algorithm predicts the neurodevelopmental outcome of preterm babies and will ultimately personalize early diagnoses of motor, language or cognitive impairment, playing a significant role in planning customized rehabilitation for each survivor.

Brain Maturation measured using deep learning and brain age 

BACKGROUND:

Dramatic brain morphological changes occur throughout the third trimester of gestation. In this study, we investigated whether the graph convolutional network (GCN) that account for cortical morphometrics and cortical surface topology as a sparse graph can predict preterm neonatal brain age. Moreover, we evaluated whether the predicted brain age (PBA) is associated with postnatal abnormalities and neurodevelopmental outcome. 

METHODS:

577 T1 MRI scans of preterm neonates from two different datasets were analyzed; the NEOCIVET pipeline generated cortical surfaces and morphological features, which were then fed to the GCN to predict the brain age globally and regionally. The brain age index (BAI; PBA minus chronological age) was used to determine the relationships among preterm birth (i.e., birthweight and birth age), perinatal brain injuries, postnatal events/clinical conditions, BAI at postnatal scan, and neurodevelopmental scores at 30 months. 

RESULTS:

GCN-based age prediction of preterm neonates without brain lesions (mean absolute error [MAE]: 0.96 weeks) outperformed conventional machine learning methods using no topological information. Structural equation models (SEM) showed that BAI mediated the influence of preterm birth and postnatal clinical factors, but not perinatal brain injuries, on neurodevelopmental outcome at 30 months of age. Regional BAI of several frontal cortices correlated with cognitive abilities at 30 months whereas BAI of left Broca’s area was associated with language functional scores.

CONCLUSIONS:

GCN provide a clinically meaningful index in measuring brain age, localizing regional growth as it relates to postnatal factors, and predicting neurodevelopmental outcome.

CONTACT

Dr. Hosung Kim (hosung.kim@loni.usc.edu)

2025 Zonal Avenue

Los Angeles, CA 90033

Report abuse
Page details
Page updated
Report abuse