Brain age estimation in subjects with anxiety/depression reliant on occipital cortex.

690 

Abstract Submission 

Owen Vega¹, Nahian Chowdhury¹, Nikhil Chaudhari^2,1, Anar Amgalan¹, Andrei Irimia^2,3,1

¹Ethel Percy Andrus Gerontology Center, Leonard Davis School of Gerontology, University of Southern California, Los Angeles, USA, ²Department of Biomedical Engineering, Viterbi School of Engineering, University of Southern California, Los Angeles, USA, ³Department of Quantitative and Computational Biology, Dana and David Dornsife College of Letters, Arts and Sciences, University of Southern California, Los Angeles, USA

Owen Vega  
Ethel Percy Andrus Gerontology Center, Leonard Davis School of Gerontology, University of Southern California
Los Angeles, USA

Nahian Chowdhury  
Ethel Percy Andrus Gerontology Center, Leonard Davis School of Gerontology, University of Southern California
Los Angeles, USA

Nikhil Chaudhari  
Department of Biomedical Engineering, Viterbi School of Engineering, University of Southern California|Ethel Percy Andrus Gerontology Center, Leonard Davis School of Gerontology, University of Southern California
Los Angeles, USA|Los Angeles, USA

Anar Amgalan, PhD  
Ethel Percy Andrus Gerontology Center, Leonard Davis School of Gerontology, University of Southern California
Los Angeles, USA

Andrei Irimia  
Department of Biomedical Engineering, Viterbi School of Engineering, University of Southern California|Department of Quantitative and Computational Biology, Dana and David Dornsife College of Letters, Arts and Sciences, University of Southern California|Ethel Percy Andrus Gerontology Center, Leonard Davis School of Gerontology, University of Southern California
Los Angeles, USA|Los Angeles, USA|Los Angeles, USA

Depression is a highly prevalent psychiatric disorder. It is the 19th most common disease in the world and the most diagnosed psychiatric disorder in adults - 1 in 5 people in the U.S. in any 6-month period suffers from depression. Depression can persist throughout life, with a 14.9-16.2% lifelong prevalence in the U.S. adult population (Kessler et al., 2015) (Richards, 2011). It can be fatal, as it increases suicide risk by 20.9-27 times, and it is a risk factor for cardiac disease (Lépine & Briley, 2011). Depression is also co-morbid with anxiety, another common psychiatric disorder. 85% of people with depression have anxiety, and 90% of people with anxiety have depression. Difficulties arise as both illnesses require different treatments (Tiller, 2013). Furthermore, the burden and prevalence of both anxiety and depression have increased following the 2020 COVID pandemic (COVID-19 Mental Disorders Collaborators, 2021). As anxiety and depression are both potentially lifelong disorders, age has been seen as a factor regarding both disorders (Gao et al., 2023). The interactions between aging and anxiety or depression remain understudied.

Our lab has developed a convolutional neural network that predicts a person's age from their T1-weighted magnetic resonance imaging (MRI) volume. The model outputs saliency maps that capture the contribution of each brain voxel to the age prediction. Here we investigate the interaction of anxiety and depression with aging by comparing patterns of saliency between subjects with these disorders and healthy controls.

Subject MRIs were obtained from the UK Biobank (UKBB). UKBB codes F32-33 and F40-43 were used to select subjects with primary or secondary diagnoses of depression and anxiety. Subjects with both anxiety and depression were excluded from the study. Subjects without MRI and subjects who had already been used to train the age prediction model were also excluded.

Sets of healthy control subjects were age-, sex-, and sample size-matched to each case group. As such, the data was separated into four groups: anxiety (N=700), anxiety-control (N=700), depression (N=940), and depression-control (N=940). Age prediction and saliency map generation were performed for each subject using the trained model. Each subject's saliency map was normalized using min-max scaling and projected onto a shared cortical surface space. Saliency maps were averaged within-group across subjects. Maps of group differences were acquired by subtracting the mean saliency of each control cohort by the mean saliency of the corresponding case cohort and dividing the difference by the average saliency of the case group.

All difference plots were smoothed using FreeSurfer 'fs_smooth.m' function with 5 iterations. Figures were plotted using MATLAB.

Saliency differences were largely symmetrical across hemispheres in both the control/anxiety (Fig. 1) and control/depression (Fig. 2) comparisons. Both anxiety and depression groups showed roughly 6% higher saliency than controls in occipital regions. Meanwhile, controls exhibited ~6% higher saliency in pre/postcentral gyri and in the central sulcus.

The two maps of saliency differences exhibit spatial overlap, suggesting that anxiety and depression influence the aging process in similar ways. Previous research has highlighted the occipital lobe as an affected area in anxiety/depression. Increased saliency at the occipital lobe in anxiety/depression subjects may relate to decreased function, which has been observed bilaterally in the occipital lobe in both task-based (Li et al., 2013) and resting-state fMRI (Peng et al., 2011) studies. A CT study also demonstrated early atrophy of the occipital lobe in subjects with depression (Tanaka et al., 1982). Grey matter volume in the right occipital region was found to be negatively correlated with anxious traits (Yin et al., 2016). Further research on how anxiety and depression influence the aging of the occipital lobe is recommended.

Psychiatric (eg. Depression, Anxiety, Schizophrenia) ¹

Aging ²

Image Registration and Computational Anatomy

Anatomy and Functional Systems

Anatomical MRI

Affective Disorders

Aging

Anxiety

Computing

Data analysis

MRI

Psychiatric Disorders

STRUCTURAL MRI