Statistical Pitfalls in Brain Age Analyses

Butler, Ellyn R.; Chen, Andrew A.; Ramadan, Rabie A.; Le, Trang T.; Ruparel, Kosha; Moore, Tyler M.; Satterthwaite, Theodore D.; Zhang, Fengqing; Shou, Haochang; Gur, Ruben; Nichols, Thomas E.; Shinohara, Russell T.

doi:10.1101/2020.06.21.163741

Cited by 10 publications

(5 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…BrainAge -as with other biological clocks such as epigenetic age measures based on methylation of the genome -is considered by some as a measure of accelerated biological aging, so that the discrepancy between a subject's true age and that estimated by the algorithm has been used as a predictor of mortality, health outcomes, and risk for disease. Even so, a recent report 24 identified a number of conceptual problems with this interpretation, noting that the brain age gap, as a residual, is dependent on age, so that any group differences on the brain age gap could simply be due to group differences in age. They further point to additional pitfalls in regressing out age from the brain age gap.…”

Section: Discussionmentioning

confidence: 99%

Accurate brain age prediction using recurrent slice-based networks

Lam

Santhalingam

Suresh

et al. 2020

Preprint

View full text Add to dashboard Cite

BrainAge (a subject’s apparent age predicted from neuroimaging data) is an important biomarker of brain aging. The deviation of BrainAge from true age has been associated with psychiatric and neurological disease, and has proven effective in predicting conversion from mild cognitive impairment (MCI) to dementia. Conventionally, 3D convolutional neural networks and their variants are used for brain age prediction. However, these networks have a larger number of parameters and take longer to train than their 2D counterparts. Here we propose a 2D slice-based recurrent neural network model, which takes in an ordered sequence of sagittal slices as input to predict the brain age. The model consists of two components: a 2D convolutional neural network (CNN), which encodes the relevant features from the slices, and a recurrent neural network (RNN) that learns the relationship between slices. We compare our method to other recently proposed methods, including 3D deep convolutional regression networks, information theoretic models, and bag-of-features (BoF) models (such as BagNet) - where the classification is based on the occurrences of local features, without taking into consideration their global spatial ordering. In our experiments, our proposed model performs comparably to, or better than, the current state of the art models, with nearly half the number of parameters and a lower convergence time.

show abstract

Section: Discussionmentioning

confidence: 99%

Accurate brain age prediction using recurrent slice-based networks

Lam

Santhalingam

Suresh

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…The role of biological sex in neuropsychiatric conditions is relevant for understanding the trajectory of brain health across the lifespan, as all of the abovementioned conditions have been linked to an "accelerated" profile of aging using neuroimaging (Hajek et al, 2019;Han et al, 2019;Koutsouleris et al, 2014;van Gestel et al, 2019) and epigenetics (Fries et al, 2017;Luo et al, 2020;Wolf et al, 2017). While these studies use algorithmic tools to determine patterns that appear consistent with "accelerated aging," it is speculative at best to conclude that biological patterns that appear older than expected for a person's chronological age are representative of true biological aging-a phenotypic concept that has no true measurement (Butler et al, 2020;Freund, 2019). Still, each of the psychiatric conditions discussed in this review has been linked to increased risk for dementia, suggesting an important role for mental distress as an etiological predictor of neurodegenerative disease.…”

Section: Sex Differences In Neurodegenerative Disordersmentioning

confidence: 99%

Sex is a defining feature of neuroimaging phenotypes in major brain disorders

Salminen

Tubi

Bright

et al. 2021

Human Brain Mapping

View full text Add to dashboard Cite

Sex is a biological variable that contributes to individual variability in brain structure and behavior. Neuroimaging studies of population-based samples have identified normative differences in brain structure between males and females, many of which are exacerbated in psychiatric and neurological conditions. Still, sex differences in MRI outcomes are understudied, particularly in clinical samples with known sex differences in disease risk, prevalence, and expression of clinical symptoms. Here we review the existing literature on sex differences in adult brain structure in normative samples and in 14 distinct psychiatric and neurological disorders. We discuss commonalities and sources of variance in study designs, analysis procedures, disease subtype effects, and the impact of these factors on MRI interpretation. Lastly, we identify key problems in the neuroimaging literature on sex differences and offer potential recommendations to address current barriers and optimize rigor and reproducibility. In particular, we emphasize the importance of large-scale neuroimaging initiatives such as the Enhancing NeuroImaging Genetics through Meta-Analyses consortium, the UK Biobank, Human Connectome Project, and others to provide unprecedented power to evaluate sex-specific phenotypes in major brain diseases.

show abstract

“…The role of biological sex in neuropsychiatric conditions is relevant for understanding the trajectory of brain health across the lifespan, as all of the abovementioned conditions have been linked to an "accelerated" profile of aging using neuroimaging [Hajek et al, 2019;Han et al, 2019;Koutsouleris et al, 2014;Van Gestel et al, 2019] and epigenetic indices [Fries et al, 2017;Luo et al, 2020;Wolf et al, 2017]. While these studies use algorithmic tools to determine patterns that appear consistent with "accelerated aging", it is speculative at best to conclude that biological patterns that appear older than expected for a person's chronological age are representative of true biological aging -a phenotypic concept that has no true measurement [Butler et al, 2020;Freund, 2019]. In the sections below we describe the most common age-related neurodegenerative conditions and evidence of sex effects on prevalence, symptoms, and neuroimaging patterns.…”

Section: Sex Differences In Age-related Neurocognitive Disordersmentioning

confidence: 99%

Sex is a defining feature of neuroimaging phenotypes in major brain disorders

Salminen¹,

Tubi²,

Bright³

et al. 2020

Preprint

View full text Add to dashboard Cite

Sex differences are found in the incidence and expression of psychiatric and neurodegenerative conditions, and many studies suggest these differences are influenced by innate biological differences between males and females and risk factors that interact with these differences. However, few studies have used neuroimaging to examine brain signatures of disease separately by sex, and many studies of sex differences have been based on small samples and their findings have not been replicated in larger cohorts. Large-scale neuroimaging initiatives such as the Enhancing NeuroImaging Genetics through Meta-Analyses (ENIGMA) consortium, the UK Biobank, Human Connectome Project, and others offer an unprecedented source of power to address important questions about the role of sex as a risk or protective factor for suboptimal brain health, as well as sex-specific neuroimaging phenotypes in brain-related illnesses. Here we review the existing neuroimaging literature on sex differences in the human brain in healthy adults and those with the most common and debilitating psychiatric and age-related neurodegenerative conditions. Finally, we discuss key gaps in this literature and opportunities of large-scale collaborative efforts to identify, characterize, and explain how biological sex influences the humanbrain in health and disease.

show abstract

Statistical Pitfalls in Brain Age Analyses

Cited by 10 publications

References 31 publications

Accurate brain age prediction using recurrent slice-based networks

Accurate brain age prediction using recurrent slice-based networks

Sex is a defining feature of neuroimaging phenotypes in major brain disorders

Sex is a defining feature of neuroimaging phenotypes in major brain disorders

Contact Info

Product

Resources

About