Biology subjects

Schulz, M.-A.

Publications and source records attributed to Schulz, M.-A..

5 recordsLinked to original sources

bioRxiv2024

Do transformers and CNNs learn different concepts of brain age?

"Predicted brain age" refers to a biomarker of structural brain health derived from machine learning analysis of T1-weighted brain magnetic resonance (MR) images. A range of machine learning methods have been used to predict brain age, with convolutional neural networks (CNNs) currently yielding state-of-the-art accuracies. Recent advances in deep learning have introduced transformers, which are conceptually distinct from CNNs, and appear to set new benchmarks in various domains of computer vision. However, transformers have not yet been applied to brain age prediction. Thus, we address two research questions: First, are transformers superior to CNNs in predicting brain age? Second, do conceptually different deep learning model architectures learn similar or different "concepts of brain age"? We adapted a Simple Vision Transformer (sViT) and a Shifted Window Transformer (SwinT) to predict brain age, and compared both models with a ResNet50 on 46,381 T1-weighted structural MR images from the UK Biobank. We found that SwinT and ResNet performed on par, while additional training samples will most likely give SwinT the edge in prediction accuracy. We identified that different model architectures may characterize different (sub-)sets of brain aging effects, representing diverging concepts of brain age. Thus, we systematically tested whether sViT, SwinT and ResNet focus on different concepts of brain age by examining variations in their predictions and clinical utility for indicating deviations in neurological and psychiatric disorders. Reassuringly, we did not find substantial differences in the structure of brain age predictions between model architectures. Based on our results, the choice of deep learning model architecture does not appear to have a confounding effect on brain age studies.

neuroscience↗

bioRxiv2024

Beyond Accuracy: Refining Brain-Age Models for Enhanced Disease Detection

This study critically reevaluates the utility of brain-age models within the context of detecting neurological and psychiatric disorders, challenging the conventional emphasis on maximizing chronological age prediction accuracy. Our analysis of T1 MRI data from 46,381 UK Biobank participants reveals that simpler machine learning models, and notably those with excessive regularization, demonstrate superior sensitivity to disease-relevant changes compared to their more complex counterparts, despite being less accurate in chronological age prediction. This counterintuitive discovery suggests that models traditionally deemed inferior might, in fact, offer a more meaningful biomarker for brain health by capturing variations pertinent to disease states. These findings challenge the assumption that accuracy-optimized brain-age models serve as useful normative models of brain aging. Optimizing for age accuracy appears misaligned with normative aims: it drives models to rely on low-variance aging features and to deemphasize higher-variance signals that are more informative about brain health and disease. Consequently, we propose a recalibration of focus towards models that, while less accurate in conventional terms, yield brain-age gaps with larger patient- control effect sizes, offering greater utility in early disease detection and understanding the multifaceted nature of brain aging.

neuroscience↗

bioRxiv2022

Comparable neural pathways link psychological stress and brain health in health and multiple sclerosis

Clinical and neuroscientific studies suggest a link between psychological stress and reduced brain health - in healthy humans and patients with neurological disorders. However, it is unclear which neural pathways mediate between stress and brain health and whether these pathways are similar in health and disease. Here, we applied an Arterial-Spin-Labeling MRI stress task in 42 healthy persons and 56 with multiple sclerosis. We tested whether brain-predicted age differences ("brain-PAD"), a highly sensitive structural brain health biomarker derived from machine learning, mirror functional connectivity between stress-responsive regions. We found that regional neural stress responsivity did not differ between groups. Although elevated brain-PAD indicated worse brain health in patients, anterior insula-occipital functional connectivity correlated with brain-PAD in both groups. Grey matter variations contributed similarly to brain-PAD in both groups. These findings suggest a generic connection between stress and brain health whose impact is amplified in multiple sclerosis by disease-specific vulnerability factors.

neuroscience↗

bioRxiv2022

Performance reserves in brain-imaging-based phenotype prediction

Machine learning studies have shown that various phenotypes can be predicted from structural and functional brain images. However, in most such studies, prediction performance ranged from moderate to disappointing. It is unclear whether prediction performance will substantially improve with larger sample sizes or whether insufficient predictive information in brain images impedes further progress. Here, we systematically assess the effect of sample size on prediction performance using sample sizes far beyond what is possible in common neuroimaging studies. We project 3-9 fold improvements in prediction performance for behavioral and mental health phenotypes when moving from one thousand to one million samples. Moreover, we find that moving from single imaging modalities to multimodal input data can lead to further improvements in prediction performance, often on par with doubling the sample size. Our analyses reveal considerable performance reserves for neuroimaging-based phenotype prediction. Machine learning models may benefit much more from extremely large neuroimaging datasets than currently believed.

neuroscience↗

bioRxiv2019

Deep learning for brains?: Different linear and nonlinear scaling in UK Biobank brain images vs. machine-learning datasets

In recent years, deep learning has unlocked unprecedented success in various domains, especially in image, text, and speech processing. These breakthroughs may hold promise for neuroscience and especially for brain-imaging investigators who start to analyze thousands of participants. However, deep learning is only beneficial if the data have nonlinear relationships and if they are exploitable at currently available sample sizes. We systematically profiled the performance of deep models, kernel models, and linear models as a function of sample size on UK Biobank brain images against established machine learning references. On MNIST and Zalando Fashion, prediction accuracy consistently improved when escalating from linear models to shallow-nonlinear models, and further improved when switching to deep-nonlinear models. The more observations were available for model training, the greater the performance gain we saw. In contrast, using structural or functional brain scans, simple linear models performed on par with more complex, highly parameterized models in age/sex prediction across increasing sample sizes. In fact, linear models kept improving as the sample size approached [~]10,000 participants. Our results indicate that the increase in performance of linear models with additional data does not saturate at the limit of current feasibility. Yet, nonlinearities of common brain scans remain largely inaccessible to both kernel and deep learning methods at any examined scale.

bioinformatics↗