Mind the gap: Performance metric evaluation in brain‐age prediction

Human Brain Mapping, Journal Year: 2022, Volume and Issue: 43(10), P. 3113 - 3129

Published: March 21, 2022

Abstract Estimating age based on neuroimaging‐derived data has become a popular approach to developing markers for brain integrity and health. While variety of machine‐learning algorithms can provide accurate predictions characteristics, there is significant variation in model accuracy reported across studies. We predicted two population‐based datasets, assessed the effects range, sample size age‐bias correction performance metrics Pearson's correlation coefficient ( r ), determination R 2 Root Mean Squared Error (RMSE) Absolute (MAE). The results showed that these vary considerably depending cohort range; values are lower when measured samples with narrower range. RMSE MAE also range due smaller errors/brain delta closer mean group. Across subsets different ranges, improve increasing size. Performance further prediction variance as well difference between training test sets, corrected indicate high accuracy—also models showing poor initial performance. In conclusion, used evaluating depend study‐specific cannot be directly compared Since generally accuracy, even poorly performing models, inspection uncorrected provides important information about underlying attributes such variance.

Language: Английский

---

Anatomically interpretable deep learning of brain age captures domain-specific cognitive impairment

Proceedings of the National Academy of Sciences, Journal Year: 2023, Volume and Issue: 120(2)

Published: Jan. 3, 2023

The gap between chronological age (CA) and biological brain age, as estimated from magnetic resonance images (MRIs), reflects how individual patterns of neuroanatomic aging deviate their typical trajectories. MRI-derived (BA) estimates are often obtained using deep learning models that may perform relatively poorly on new data or lack interpretability. This study introduces a convolutional neural network (CNN) to estimate BA after training the MRIs 4,681 cognitively normal (CN) participants testing 1,170 CN an independent sample. estimation errors notably lower than those previous studies. At both cohort levels, CNN provides detailed anatomic maps reveal sex dimorphisms neurocognitive trajectories in adults with mild cognitive impairment (MCI, N = 351) Alzheimer’s disease (AD, 359). In individuals MCI (54% whom were diagnosed dementia within 10.9 y MRI acquisition), is significantly better CA capturing symptom severity, functional disability, executive function. Profiles dimorphism lateralization also map onto change reflect decline. Significant associations measures suggest proposed framework can map, systematically, relationship aging-related neuroanatomy changes AD. Early identification such help screen according AD risk.

Language: Английский

---

The genetic architecture of multimodal human brain age

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: March 23, 2024

Abstract The complex biological mechanisms underlying human brain aging remain incompletely understood. This study investigated the genetic architecture of three age gaps (BAG) derived from gray matter volume (GM-BAG), white microstructure (WM-BAG), and functional connectivity (FC-BAG). We identified sixteen genomic loci that reached genome-wide significance (P-value < 5×10 −8 ). A gene-drug-disease network highlighted genes linked to GM-BAG for treating neurodegenerative neuropsychiatric disorders WM-BAG cancer therapy. displayed most pronounced heritability enrichment in variants within conserved regions. Oligodendrocytes astrocytes, but not neurons, exhibited notable WM FC-BAG, respectively. Mendelian randomization potential causal effects several chronic diseases on aging, such as type 2 diabetes AD WM-BAG. Our results provide insights into genetics with clinical implications lifestyle therapeutic interventions. All are publicly available at https://labs.loni.usc.edu/medicine .

Language: Английский

---

Advances in Neuroimaging and Deep Learning for Emotion Detection: A Systematic Review of Cognitive Neuroscience and Algorithmic Innovations

Diagnostics, Journal Year: 2025, Volume and Issue: 15(4), P. 456 - 456

Published: Feb. 13, 2025

Background/Objectives: The following systematic review integrates neuroimaging techniques with deep learning approaches concerning emotion detection. It, therefore, aims to merge cognitive neuroscience insights advanced algorithmic methods in pursuit of an enhanced understanding and applications recognition. Methods: study was conducted PRISMA guidelines, involving a rigorous selection process that resulted the inclusion 64 empirical studies explore modalities such as fMRI, EEG, MEG, discussing their capabilities limitations It further evaluates architectures, including neural networks, CNNs, GANs, terms roles classifying emotions from various domains: human-computer interaction, mental health, marketing, more. Ethical practical challenges implementing these systems are also analyzed. Results: identifies fMRI powerful but resource-intensive modality, while EEG MEG more accessible high temporal resolution limited by spatial accuracy. Deep models, especially CNNs have performed well emotions, though they do not always require large diverse datasets. Combining data behavioral features improves classification performance. However, ethical challenges, privacy bias, remain significant concerns. Conclusions: has emphasized efficiencies detection, technical were highlighted. Future research should integrate advances, establish innovative enhance system reliability applicability.

Language: Английский

---

Linking brain maturation and puberty during early adolescence using longitudinal brain age prediction in the ABCD cohort

Developmental Cognitive Neuroscience, Journal Year: 2023, Volume and Issue: 60, P. 101220 - 101220

Published: Feb. 22, 2023

The temporal characteristics of adolescent neurodevelopment are shaped by a complex interplay genetic, biological, and environmental factors. Using large longitudinal dataset children aged 9–13 from the Adolescent Brain Cognitive Development (ABCD) study we tested associations between pubertal status brain maturation. maturation was assessed using age prediction based on convolutional neural networks minimally processed T1-weighted structural MRI data. provided highly accurate reliable estimates individual age, with an overall mean absolute error 0.7 1.4 years at two timepoints respectively, intraclass correlation 0.65. Linear mixed effects (LME) models accounting for sex showed that average, one unit increase in maturational level associated 2.22 months higher across time points (β = 0.10, p < .001). Moreover, annualized change development weakly related to rate .047, 0.04). These results demonstrate link sexual early adolescence, provides basis further investigations sociobiological impacts puberty life outcomes.

Language: Английский

---

Prediction of brain age using structural magnetic resonance imaging: A comparison of accuracy and test–retest reliability of publicly available software packages

Human Brain Mapping, Journal Year: 2023, Volume and Issue: 44(17), P. 6139 - 6148

Published: Oct. 16, 2023

Brain age prediction algorithms using structural magnetic resonance imaging (MRI) aim to assess the biological of human brain. The difference between a person's chronological and estimated brain is thought reflect deviations from normal aging trajectory, indicating slower or accelerated process. Several pre-trained software packages for predicting are publicly available. In this study, we perform comparison such with respect (1) predictive accuracy, (2) test-retest reliability, (3) ability track progression over time. We evaluated six packages: brainageR, DeepBrainNet, brainage, ENIGMA, pyment, mccqrnn. accuracy reliability were assessed on MRI data 372 healthy people aged 18.4 86.2 years (mean 38.7 ± 17.5 years). All showed significant correlations predicted (r = 0.66-0.97, p < 0.001), pyment displaying strongest correlation. mean absolute error was 3.56 (pyment) 9.54 (ENIGMA). mccqrnn superior in terms (ICC values 0.94-0.98), as well longer time span. Of packages, brainageR consistently highest reliability.

Language: Английский

---

Brain asymmetries from mid- to late life and hemispheric brain age

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: Feb. 1, 2024

Abstract The human brain demonstrates structural and functional asymmetries which have implications for ageing mental neurological disease development. We used a set of magnetic resonance imaging (MRI) metrics derived from diffusion MRI data in N =48,040 UK Biobank participants to evaluate age-related differences asymmetry. Most regional grey white matter presented asymmetry, were higher later life. Informed by these results, we conducted hemispheric age (HBA) predictions left/right multimodal metrics. HBA was concordant conventional predictions, using both hemispheres, but offers supplemental general marker asymmetry when setting into relationship with each other. In contrast WM asymmetries, discrepancies are lower at ages. Our findings outline various sex-specific differences, particularly important estimates, the value further investigating role

Language: Английский

---

Brain structure ages—A new biomarker for multi‐disease classification

Human Brain Mapping, Journal Year: 2024, Volume and Issue: 45(1)

Published: Jan. 1, 2024

Age is an important variable to describe the expected brain's anatomy status across normal aging trajectory. The deviation from that normative trajectory may provide some insights into neurological diseases. In neuroimaging, predicted brain age widely used analyze different However, using only gap information (i.e., difference between chronological and estimated age) can be not enough informative for disease classification problems. this paper, we propose extend notion of global by estimating structure ages structural magnetic resonance imaging. To end, ensemble deep learning models first estimate a 3D map voxel-wise estimation). Then, segmentation mask obtain final ages. This biomarker in several situations. First, it enables accurately purpose anomaly detection at population level. situation, our approach outperforms state-of-the-art methods. Second, compute process each structure. feature multi-disease task accurate differential diagnosis subject Finally, deviations individuals visualized, providing about abnormality helping clinicians real medical contexts.

Language: Английский

---

Deep learning to quantify the pace of brain aging in relation to neurocognitive changes

Proceedings of the National Academy of Sciences, Journal Year: 2025, Volume and Issue: 122(10)

Published: Feb. 24, 2025

Brain age (BA), distinct from chronological (CA), can be estimated MRIs to evaluate neuroanatomic aging in cognitively normal (CN) individuals. BA, however, is a cross-sectional measure that summarizes cumulative since birth. Thus, it conveys poorly recent or contemporaneous trends, which better quantified by the (temporal) pace P of brain aging. Many approaches map , rely on quantifying DNA methylation whole-blood cells, blood–brain barrier separates neural cells. We introduce three-dimensional convolutional network (3D-CNN) estimate noninvasively longitudinal MRI. Our model (LM) trained 2,055 CN adults, validated 1,304 and further applied an independent cohort 104 adults 140 patients with Alzheimer’s disease (AD). In its test set, LM computes mean absolute error (MAE) 0.16 y (7% error). This significantly outperforms most accurate model, whose MAE 1.85 has 83% error. By synergizing interpretable CNN saliency approach, we anatomic variations regional rates differ according sex, decade life, neurocognitive status. estimates are associated changes cognitive functioning across domains. underscores LM’s ability way captures relationship between research complements existing strategies for AD risk assessment individuals’ adverse change age.

Language: Английский

---

Genetically supported targets and drug repurposing for brain aging: A systematic study in the UK Biobank

Science Advances, Journal Year: 2025, Volume and Issue: 11(11)

Published: March 12, 2025

Brain age gap (BAG), the deviation between estimated brain and chronological age, is a promising marker of health. However, genetic architecture reliable targets for aging remains poorly understood. In this study, we estimate magnetic resonance imaging (MRI)–based using deep learning models trained on UK Biobank validated with three external datasets. A genome-wide association study BAG identified two unreported loci seven previously reported loci. By integrating Mendelian Randomization (MR) colocalization analysis eQTL pQTL data, prioritized genetically supported druggable genes, including MAPT , TNFSF12 GZMB SIRPB1 GNLY NMB C1RL as aging. We rediscovered 13 potential drugs evidence from clinical trials several strong support. Our provides insights into basis aging, potentially facilitating drug development to extend health span.

Language: Английский

---