Invertible Modeling of Bidirectional Relationships in Neuroimaging With Normalizing Flows: Application to Brain Aging

Analysis and visualization of the effect of multiple sclerosis on biological brain age

Introduction

The rate of neurodegeneration in multiple sclerosis (MS) is an important biomarker for disease progression but can be challenging to quantify. The brain age gap, which quantifies the difference between a patient's chronological and their estimated biological brain age, might be a valuable biomarker of neurodegeneration in patients with MS. Thus, the aim of this study was to investigate the value of an image-based prediction of the brain age gap using a deep learning model and compare brain age gap values between healthy individuals and patients with MS.

Methods

A multi-center dataset consisting of 5,294 T1-weighted magnetic resonance images of the brain from healthy individuals aged between 19 and 89 years was used to train a convolutional neural network (CNN) for biological brain age prediction. The trained model was then used to calculate the brain age gap in 195 patients with relapsing remitting MS (20-60 years). Additionally, saliency maps were generated for healthy subjects and patients with MS to identify brain regions that were deemed important for the brain age prediction task by the CNN.

Results

Overall, the application of the CNN revealed accelerated brain aging with a larger brain age gap for patients with MS with a mean of 6.98 ± 7.18 years in comparison to healthy test set subjects (0.23 ± 4.64 years). The brain age gap for MS patients was weakly to moderately correlated with age at disease onset (ρ = -0.299, p < 0.0001), EDSS score (ρ = 0.206, p = 0.004), disease duration (ρ = 0.162, p = 0.024), lesion volume (ρ = 0.630, p < 0.0001), and brain parenchymal fraction (ρ = -0.718, p < 0.0001). The saliency maps indicated significant differences in the lateral ventricle (p < 0.0001), insula (p < 0.0001), third ventricle (p < 0.0001), and fourth ventricle (p = 0.0001) in the right hemisphere. In the left hemisphere, the inferior lateral ventricle (p < 0.0001) and the third ventricle (p < 0.0001) showed significant differences. Furthermore, the Dice similarity coefficient showed the highest overlap of salient regions between the MS patients and the oldest healthy subjects, indicating that neurodegeneration is accelerated in this patient cohort.

Discussion

In conclusion, the results of this study show that the brain age gap is a valuable surrogate biomarker to measure disease progression in patients with multiple sclerosis.

A machine learning approach using conditional normalizing flow to address extreme class imbalance problems in personal health records

BACKGROUND

Supervised machine learning models have been widely used to predict and get insight into diseases by classifying patients based on personal health records. However, a class imbalance is an obstacle that disrupts the training of the models. In this study, we aimed to address class imbalance with a conditional normalizing flow model, one of the deep-learning-based semi-supervised models for anomaly detection. It is the first introduction of the normalizing flow algorithm for tabular biomedical data.

METHODS

We collected personal health records from South Korean citizens (n = 706), featuring genetic data obtained from direct-to-customer service (microarray chip), medical health check-ups, and lifestyle log data. Based on the health check-up data, six chronic diseases were labeled (obesity, diabetes, hypertriglyceridemia, dyslipidemia, liver dysfunction, and hypertension). After preprocessing, supervised classification models and semi-supervised anomaly detection models, including conditional normalizing flow, were evaluated for the classification of diabetes, which had extreme target imbalance (about 2%), based on AUROC and AUPRC. In addition, we evaluated their performance under the assumption of insufficient collection for patients with other chronic diseases by undersampling disease-affected samples.

RESULTS

While LightGBM (the best-performing model among supervised classification models) showed AUPRC 0.16 and AUROC 0.82, conditional normalizing flow achieved AUPRC 0.34 and AUROC 0.83 during fifty evaluations of the classification of diabetes, whose base rate was very low, at 0.02. Moreover, conditional normalizing flow performed better than the supervised model under a few disease-affected data numbers for the other five chronic diseases - obesity, hypertriglyceridemia, dyslipidemia, liver dysfunction, and hypertension. For example, while LightGBM performed AUPRC 0.20 and AUROC 0.75, conditional normalizing flow showed AUPRC 0.30 and AUROC 0.74 when predicting obesity, while undersampling disease-affected samples (positive undersampling) lowered the base rate to 0.02.

CONCLUSIONS

Our research suggests the utility of conditional normalizing flow, particularly when the available cases are limited, for predicting chronic diseases using personal health records. This approach offers an effective solution to deal with sparse data and extreme class imbalances commonly encountered in the biomedical context.

Identifying Biases in a Multicenter MRI Database for Parkinson's Disease Classification: Is the Disease Classifier a Secret Site Classifier?

Sharing multicenter imaging datasets can be advantageous to increase data diversity and size but may lead to spurious correlations between site-related biological and non-biological image features and target labels, which machine learning (ML) models may exploit as shortcuts. To date, studies analyzing how and if deep learning models may use such effects as a shortcut are scarce. Thus, the aim of this work was to investigate if site-related effects are encoded in the feature space of an established deep learning model designed for Parkinson's disease (PD) classification based on T1-weighted MRI datasets. Therefore, all layers of the PD classifier were frozen, except for the last layer of the network, which was replaced by a linear layer that was exclusively re-trained to predict three potential bias types (biological sex, scanner type, and originating site). Our findings based on a large database consisting of 1880 MRI scans collected across 41 centers show that the feature space of the established PD model (74% accuracy) can be used to classify sex (75% accuracy), scanner type (79% accuracy), and site location (71% accuracy) with high accuracies despite this information never being explicitly provided to the PD model during original training. Overall, the results of this study suggest that trained image-based classifiers may use unwanted shortcuts that are not meaningful for the actual clinical task at hand. This finding may explain why many image-based deep learning models do not perform well when applied to data from centers not contributing to the training set.

Providing clinical context to the spatio-temporal analysis of 4D CT perfusion to predict acute ischemic stroke lesion outcomes

Acute ischemic stroke is a leading cause of mortality and morbidity worldwide. Timely identification of the extent of a stroke is crucial for effective treatment, whereas spatio-temporal (4D) Computed Tomography Perfusion (CTP) imaging is playing a critical role in this process. Recently, the first deep learning-based methods that leverage the full spatio-temporal nature of perfusion imaging for predicting stroke lesion outcomes have been proposed. However, clinical information is typically not integrated into the learning process, which may be helpful to improve the tissue outcome prediction given the known influence of various factors (i.e., physiological, demographic, and treatment factors) on lesion growth. Cross-attention, a multimodal fusion strategy, has been successfully used to combine information from multiple sources, but it has yet to be applied to stroke lesion outcome prediction. Therefore, this work aimed to develop and evaluate a novel multimodal and spatio-temporal deep learning model that utilizes cross-attention to combine information from 4D CTP and clinical metadata simultaneously to predict stroke lesion outcomes. The proposed model was evaluated using a dataset of 70 acute ischemic stroke patients, demonstrating significantly improved volume estimates (mean error = 19 ml) compared to a baseline unimodal approach (mean error = 35 ml, p< 0.05). The proposed model allows generating attention maps and counterfactual outcome scenarios to investigate the relevance of clinical variables in predicting stroke lesion outcomes at a patient level, helping to provide a better understanding of the model's decision-making process.

Multimodal brain age prediction fusing morphometric and imaging data and association with cardiovascular risk factors

Introduction

The difference between the chronological and biological brain age, called the brain age gap (BAG), has been identified as a promising biomarker to detect deviation from normal brain aging and to indicate the presence of neurodegenerative diseases. Moreover, the BAG has been shown to encode biological information about general health, which can be measured through cardiovascular risk factors. Current approaches for biological brain age estimation, and therefore BAG estimation, either depend on hand-crafted, morphological measurements extracted from brain magnetic resonance imaging (MRI) or on direct analysis of brain MRI images. The former can be processed with traditional machine learning models while the latter is commonly processed with convolutional neural networks (CNNs). Using a multimodal setting, this study aims to compare both approaches in terms of biological brain age prediction accuracy and biological information captured in the BAG.

Methods

T1-weighted MRI, containing brain tissue information, and magnetic resonance angiography (MRA), providing information about brain arteries, from 1,658 predominantly healthy adults were used. The volumes, surface areas, and cortical thickness of brain structures were extracted from the T1-weighted MRI data, while artery density and thickness within the major blood flow territories and thickness of the major arteries were extracted from MRA data. Independent multilayer perceptron and CNN models were trained to estimate the brain age from the hand-crafted features and image data, respectively. Next, both approaches were fused to assess the benefits of combining image data and hand-crafted features for brain age prediction.

Results

The combined model achieved a mean absolute error of 4 years between the chronological and predicted biological brain age. Among the independent models, the lowest mean absolute error was observed for the CNN using T1-weighted MRI data (4.2 years). When evaluating the BAGs obtained using the different approaches and imaging modalities, diverging associations between cardiovascular risk factors were found. For example, BAGs obtained from the CNN models showed an association with systolic blood pressure, while BAGs obtained from hand-crafted measurements showed greater associations with obesity markers.

Discussion

In conclusion, the use of more diverse sources of data can improve brain age estimation modeling and capture more diverse biological deviations from normal aging.

An analysis of the effects of limited training data in distributed learning scenarios for brain age prediction

OBJECTIVE

Distributed learning avoids problems associated with central data collection by training models locally at each site. This can be achieved by federated learning (FL) aggregating multiple models that were trained in parallel or training a single model visiting sites sequentially, the traveling model (TM). While both approaches have been applied to medical imaging tasks, their performance in limited local data scenarios remains unknown. In this study, we specifically analyze FL and TM performances when very small sample sizes are available per site.

MATERIALS AND METHODS

2025 T1-weighted magnetic resonance imaging scans were used to investigate the effect of sample sizes on FL and TM for brain age prediction. We evaluated models across 18 scenarios varying the number of samples per site (1, 2, 5, 10, and 20) and the number of training rounds (20, 40, and 200).

RESULTS

Our results demonstrate that the TM outperforms FL, for every sample size examined. In the extreme case when each site provided only one sample, FL achieved a mean absolute error (MAE) of 18.9 ± 0.13 years, while the TM achieved a MAE of 6.21 ± 0.50 years, comparable to central learning (MAE = 5.99 years).

DISCUSSION

Although FL is more commonly used, our study demonstrates that TM is the best implementation for small sample sizes.

CONCLUSION

The TM offers new opportunities to apply machine learning models in rare diseases and pediatric research but also allows even small hospitals to contribute small datasets.