Deep Learning and Multiplex Networks for Accurate Modeling of Brain Age

Amoroso, Nicola; Rocca, Marianna La; Bellantuono, Loredana; Diacono, Domenico; Fanizzi, Annarita; Lella, Eufemia; Lombardi, Angela; Maggipinto, T.; Tangaro, Sabina; Bellotti, R.

doi:10.3389/fnagi.2019.00115

Cited by 48 publications

(32 citation statements)

References 57 publications

(61 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We used linear models in order to facilitate better interpretations. More sophisticated yet complicated models could also have been used to possibly have better predictive power, for example, to have higher prediction accuracy in age prediction [116]. Nevertheless, using a simple linear predictor, we were able to achieve a substantial cross-validated accuracy, which effectively demonstrates the capability of our model in capturing the brain maturation patterns.…”

Section: Limitationsmentioning

confidence: 84%

Deviation from normative brain development is associated with symptom severity in autism spectrum disorder

et al. 2019

View full text Add to dashboard Cite

Background: Autism spectrum disorder (ASD) is a heterogeneous neurodevelopmental condition. The degree to which the brain development in ASD deviates from typical brain development, and how this deviation relates to observed behavioral outcomes at the individual level are not well-studied. We hypothesize that the degree of deviation from typical brain development of an individual with ASD would relate to observed symptom severity. Methods: The developmental changes in anatomical (cortical thickness, surface area, and volume) and diffusion metrics (fractional anisotropy and apparent diffusion coefficient) were compared between a sample of ASD (n = 247) and typically developing children (TDC) (n = 220) aged 6-25. Machine learning was used to predict age (brain age) from these metrics in the TDC sample, to define a normative model of brain development. This model was then used to compute brain age in the ASD sample. The difference between chronological age and brain age was considered a developmental deviation index (DDI), which was then correlated with ASD symptom severity.Results: Machine learning model trained on all five metrics accurately predicted age in the TDC (r = 0.88) and the ASD (r = 0.85) samples, with dominant contributions to the model from the diffusion metrics. Within the ASD group, the DDI derived from fractional anisotropy was correlated with ASD symptom severity (r = − 0.2), such that individuals with the most advanced brain age showing the lowest severity, and individuals with the most delayed brain age showing the highest severity. Limitations: This work investigated only linear relationships between five specific brain metrics and only one measure of ASD symptom severity in a limited age range. Reported effect sizes are moderate. Further work is needed to investigate developmental differences in other age ranges, other aspects of behavior, other neurobiological measures, and in an independent sample before results can be clinically applicable. Conclusions: Findings demonstrate that the degree of deviation from typical brain development relates to ASD symptom severity, partially accounting for the observed heterogeneity in ASD. Our approach enables characterization of each individual with reference to normative brain development and identification of distinct developmental subtypes, facilitating a better understanding of developmental heterogeneity in ASD.

show abstract

Section: Limitationsmentioning

confidence: 84%

Deviation from normative brain development is associated with symptom severity in autism spectrum disorder

et al. 2019

View full text Add to dashboard Cite

show abstract

“…The prediction accuracy that we obtained is in the same range as has been demonstrated by previous studies while using different morphological measures and age prediction methods (

) [ 22 , 51 , 52 , 53 , 54 , 55 ]. Most of these studies are focused on younger subjects (age

years) and reported MAE

[ 51 , 52 , 54 , 55 ], while other works showed that the prediction error increases with increasing age with MAE

in the same age range of our analysis [ 22 , 53 , 56 ]. Our results seem to confirm such an observation: even though the overall mean prediction errors resulting from random forest age regression for the age covariate harmonization and no harmonization are similar to those reported in literature, all of the models show worse performance for age

years.…”

Section: Discussionmentioning

confidence: 98%

Extensive Evaluation of Morphological Statistical Harmonization for Brain Age Prediction

et al. 2020

Self Cite

View full text Add to dashboard Cite

Characterizing both neurodevelopmental and aging brain structural trajectories is important for understanding normal biological processes and atypical patterns that are related to pathological phenomena. Initiatives to share open access morphological data contributed significantly to the advance in brain structure characterization. Indeed, such initiatives allow large brain morphology multi-site datasets to be shared, which increases the statistical sensitivity of the outcomes. However, using neuroimaging data from multi-site studies requires harmonizing data across the site to avoid bias. In this work we evaluated three different harmonization techniques on the Autism Brain Imaging Data Exchange (ABIDE) dataset for age prediction analysis in two groups of subjects (i.e., controls and autism spectrum disorder). We extracted the morphological features from T1-weighted images of a mixed cohort of 654 subjects acquired from 17 sites to predict the biological age of the subjects using three machine learning regression models. A machine learning framework was developed to quantify the effects of the different harmonization strategies on the final performance of the models and on the set of morphological features that are relevant to the age prediction problem in both the presence and absence of pathology. The results show that, even if two harmonization strategies exhibit similar accuracy of predictive models, a greater mismatch occurs between the sets of most age-related predictive regions for the Autism Spectrum Disorder (ASD) subjects. Thus, we propose to use a stability index to extract meaningful features for a robust clinical validation of the outcomes of multiple harmonization strategies.

show abstract

“…The success of CNNs in computer vision has led to numerous applications in medical imaging and more recently in age prediction from neuroimaging data (17,(45)(46)(47)(48)(49).…”

Section: Model 3: Six-layer Convolutional Neural Networkmentioning

confidence: 99%

Ensemble Learning of Convolutional Neural Network, Support Vector Machine, and Best Linear Unbiased Predictor for Brain Age Prediction: ARAMIS Contribution to the Predictive Analytics Competition 2019 Challenge

et al. 2020

View full text Add to dashboard Cite

We ranked third in the Predictive Analytics Competition (PAC) 2019 challenge by achieving a mean absolute error (MAE) of 3.33 years in predicting age from T1-weighted MRI brain images. Our approach combined seven algorithms that allow generating predictions when the number of features exceeds the number of observations, in particular, two versions of best linear unbiased predictor (BLUP), support vector machine (SVM), two shallow convolutional neural networks (CNNs), and the famous ResNet and Inception V1. Ensemble learning was derived from estimating weights via linear regression in a hold-out subset of the training sample. We further evaluated and identified factors that could influence prediction accuracy: choice of algorithm, ensemble learning, and features used as input/MRI image processing. Our prediction error was correlated with age, and absolute error was greater for older participants, suggesting to increase the training sample for this subgroup. Our results may be used to guide researchers to build age predictors on healthy individuals, which can be used in research and in the clinics as non-specific predictors of disease status.

show abstract

Deep Learning and Multiplex Networks for Accurate Modeling of Brain Age

Cited by 48 publications

References 57 publications

Deviation from normative brain development is associated with symptom severity in autism spectrum disorder

Deviation from normative brain development is associated with symptom severity in autism spectrum disorder

Extensive Evaluation of Morphological Statistical Harmonization for Brain Age Prediction

Ensemble Learning of Convolutional Neural Network, Support Vector Machine, and Best Linear Unbiased Predictor for Brain Age Prediction: ARAMIS Contribution to the Predictive Analytics Competition 2019 Challenge

Contact Info

Product

Resources

About