Bias-adjustment in neuroimaging-based brain age frameworks: A robust scheme

Beheshti, Iman; Nugent, Scott; Potvin, Olivier; Duchesne, Simon

doi:10.1016/j.nicl.2019.102063

Cited by 145 publications

(155 citation statements)

References 20 publications

Supporting

Mentioning

148

Contrasting

Order By: Relevance

“…To mitigate the residuals' dependence on age, researchers apply the following algorithm (Beheshti et al, 2019;Smith et al, 2019;Liang et al, 2019;Chung et al, 2018) Then, for a left out subject i with brain data B i1 , B i2 , . .…”

Section: Risks Of Using a Modified Brain Age Gapmentioning

confidence: 99%

“…(7). If Beheshti et al (2019) implemented their method as stated, they would have found that the correlation after their adjustment dramatically underestimates of the true correlation (see Figure 2). To illustrate what correlation would be obtained if the reported transformations were used, correlations using both the transformation and the equivalent function are plotted.…”

Section: Acknowledgementsmentioning

confidence: 99%

“…Adding and subtracting MBAG from age results in the same in-flated correlation with ageModitified predicted age has been calculated in the literature as either MBAG i = A i −BAG i or MBAG i = A i + BAG i . In both cases, the main results from the paper applies since the correlation between age and Correlation as a result ofBeheshti et al (2019)'s reported transformation is a function of the true correlation. The Beheshti correlation is the correlation between age and predicted age minus predicted BAG.…”

mentioning

confidence: 94%

“…It is important to note that regression towards the mean is not a failure, but a feature, of regression and related methods. If there is any randomness in a process, observations will tend towards the mean of the outcome variable rather than remain as extreme as they were upon initial sampling (Stigler, 1997 (Beheshti, Nugent, Potvin, & Duchesne, 2019;Smith et al, 2019;Liang et al, 2019;Le et al, 2018).…”

Section: Introductionmentioning

confidence: 99%

“…Beheshti et al (2019) correlationBeheshti et al (2019) suggest subtractingα +γA i fromÂ i , and calling this new value the corrected predicted age: Corr(A,Â − (α +γA)) = Corr(A,Â −γA) = Cov(A,Â −γA) Var(A) × Var(Â −γA) = −γVar(A) + Cov(A,Â) Var(A) × Var(Â) +γ 2 Var(A) − 2γCov(A,Â)…”

mentioning

confidence: 99%

See 4 more Smart Citations

Statistical Pitfalls in Brain Age Analyses

Butler

Chen

Ramadan

et al. 2020

Preprint

View full text Add to dashboard Cite

Over the past decade, there has been an abundance of research on the difference between age and age predicted using brain features, which is commonly referred to as the "brain age gap". Researchers have identified that the brain age gap, as a residual, is dependent on age. As such, any group differences on the brain age gap could simply be due to group differences on age. To mitigate the brain age gap's dependence on age, it has been proposed that age be regressed out of the brain age gap. If this modified brain age gap (MBAG) is treated as a corrected deviation from age, model accuracy statistics such as R 2 will be artificially inflated. Given the limitations of proposed brain age analyses, further theoretical work is warranted to determine the best way to quantify deviation from normality.

show abstract

Section: Risks Of Using a Modified Brain Age Gapmentioning

confidence: 99%

Section: Acknowledgementsmentioning

confidence: 99%

mentioning

confidence: 94%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Statistical Pitfalls in Brain Age Analyses

Butler

Chen

Ramadan

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

Successful cognitive aging is associated with thicker anterior cingulate cortex and lower tau deposition compared to typical aging

Pezzoli,

Giorgio,

Martersteck

et al. 2023

Alzheimer's & Dementia

View full text Add to dashboard Cite

INTRODUCTIONThere is no consensus on either the definition of successful cognitive aging (SA) or the underlying neural mechanisms.METHODSWe examined the agreement between new and existing definitions using: (1) a novel measure, the cognitive age gap (SA‐CAG, cognitive‐predicted age minus chronological age), (2) composite scores for episodic memory (SA‐EM), (3) non‐memory cognition (SA‐NM), and (4) the California Verbal Learning Test (SA‐CVLT).RESULTSFair to moderate strength of agreement was found between the four definitions. Most SA groups showed greater cortical thickness compared to typical aging (TA), especially in the anterior cingulate and midcingulate cortices and medial temporal lobes. Greater hippocampal volume was found in all SA groups except SA‐NM. Lower entorhinal 18F‐Flortaucipir (FTP) uptake was found in all SA groups.DISCUSSIONThese findings suggest that a feature of SA, regardless of its exact definition, is resistance to tau pathology and preserved cortical integrity, especially in the anterior cingulate and midcingulate cortices.Highlights Different approaches have been used to define successful cognitive aging (SA). Regardless of definition, different SA groups have similar brain features. SA individuals have greater anterior cingulate thickness and hippocampal volume. Lower entorhinal tau deposition, but not amyloid beta is related to SA. A combination of cortical integrity and resistance to tau may be features of SA.

show abstract

Brain age prediction: A comparison between machine learning models using region‐ and voxel‐based morphometric data

Baecker

Dafflon

Costa

et al. 2021

Human Brain Mapping

View full text Add to dashboard Cite

Brain morphology varies across the ageing trajectory and the prediction of a person's age using brain features can aid the detection of abnormalities in the ageing process.Existing studies on such "brain age prediction" vary widely in terms of their methods and type of data, so at present the most accurate and generalisable methodological approach is unclear. Therefore, we used the UK Biobank data set (N = 10,824, age range 47-73) to compare the performance of the machine learning models support vector regression, relevance vector regression and Gaussian process regression on whole-brain region-based or voxel-based structural magnetic resonance imaging data with or without dimensionality reduction through principal component analysis. Performance was assessed in the validation set through cross-validation as well as an independent test set. The models achieved mean absolute errors between 3.7 and 4.7 years, with those trained on voxel-level data with principal component analysis performing best. Overall, we observed little difference in performance between models trained on the same data type, indicating that the type of input data had greater impact on performance than model choice. All code is provided online in the hope that this will aid future research.

show abstract

Bias-adjustment in neuroimaging-based brain age frameworks: A robust scheme

Cited by 145 publications

References 20 publications

Statistical Pitfalls in Brain Age Analyses

Statistical Pitfalls in Brain Age Analyses

Successful cognitive aging is associated with thicker anterior cingulate cortex and lower tau deposition compared to typical aging

Brain age prediction: A comparison between machine learning models using region‐ and voxel‐based morphometric data

Contact Info

Product

Resources

About