Resting‐state functional magnetic resonance imaging (rs‐fMRI) helps characterize regional interactions that occur in the human brain at a resting state. Existing research often attempts to explore fMRI biomarkers that best predict brain disease progression using machine/deep learning techniques. Previous fMRI studies have shown that learning‐based methods usually require a large amount of labeled training data, limiting their utility in clinical practice where annotating data is often time‐consuming and labor‐intensive. To this end, we propose an unsupervised contrastive graph learning (UCGL) framework for fMRI‐based brain disease analysis, in which a pretext model is designed to generate informative fMRI representations using unlabeled training data, followed by model fine‐tuning to perform downstream disease identification tasks. Specifically, in the pretext model, we first design a bi‐level fMRI augmentation strategy to increase the sample size by augmenting blood‐oxygen‐level‐dependent (BOLD) signals, and then employ two parallel graph convolutional networks for fMRI feature extraction in an unsupervised contrastive learning manner. This pretext model can be optimized on large‐scale fMRI datasets, without requiring labeled training data. This model is further fine‐tuned on to‐be‐analyzed fMRI data for downstream disease detection in a task‐oriented learning manner. We evaluate the proposed method on three rs‐fMRI datasets for cross‐site and cross‐dataset learning tasks. Experimental results suggest that the UCGL outperforms several state‐of‐the‐art approaches in automated diagnosis of three brain diseases (i.e., major depressive disorder, autism spectrum disorder, and Alzheimer's disease) with rs‐fMRI data.
31 October 2019 in IEEE Transactions on Biomedical Engineering
Functional connectivity network (FCN) has become an increasingly important approach to gain a better understanding of the brain, as well as discover informative biomarkers for diagnosis of neurodegenerative diseases. Due to its importance, many FCN estimation methods have been developed in the past decades, including methods based on the classical Pearson's correlation, (regularized) partial correlation, and some higher-order variants. However, most of the existing methods estimate one FCN at a time, thus ignoring the possibly shared structure among FCNs from different subjects. Recently, researchers introduce group constraints (or population priors) into FCN estimation by assuming that FCNs are topologically identical across subjects. Obviously, such a constraint/prior is too strong to be satisfied in practice, especially when both patients and healthy subjects are involved simultaneously in the group. To address this problem, we propose a novel FCN estimation approach based on an assumption that the involved FCNs have similar but not necessarily identical topology. More specifically, we implement this idea under a two-step learning framework. First, we independently estimate FCNs based on traditional methods, such as Pearson's correltion and sparse representation, making sure that each FCN captures the specific properties of the corresponding subject. Then, we stack the estimated FCNs (in fact, their adjacency matrices) into a tensor, and refine the stacked FCNs via low-rank tensor approximation. Finally, we apply the improved FCNs to identify subjects with mild cognitive impairment (MCI) from healthy controls, and achieve a higher classification accuracy.
As one of the most common neurological disorders, epilepsy causes great physical and psychological damage to the patients. The long-term recurrent and unprovoked seizures make the prediction necessary. In this paper, a novel approach for epileptic seizure prediction based on successive variational mode decomposition (SVMD) and transformers is proposed. SVMD is extended to multidimensional form for time-frequency analysis of multi-channel signals. It could adaptively extract common band-limited intrinsic modes among all channels on different time scales by solving a variational optimization problem. In the proposed seizure prediction method, data are first decomposed into multiple modes on different time scales by multivariate SVMD, and then, irrelevant modes are removed for preprocessing. Finally, power spectrum of denoised data is input to a pre-trained bidirectional encoder representations from transformers (BERTs) for prediction. The BERT could identify the mode information related to epileptic seizures in time-frequency domain. It shows fair prediction performance on an intracranial EEG dataset with the average sensitivity of 0.86 and FPR of 0.18/h.
Brain functional network (BFN) analysis is becoming a crucial way to explore the inherent organized pattern of the brain and reveal potential biomarkers for diagnosing neurological or psychological disorders. In so doing, a well-estimated BFN is of great concern. In practice, however, noises or artifacts involved in the observed data (i.e., fMRI time series in this paper) generally lead to a poor estimation of BFN, and thus a complex preprocessing pipeline is often used to improve the quality of the data prior to BFN estimation. One of the popular preprocessing steps is data-scrubbing that aims at removing “bad” volumes from the fMRI time series according to the amplitude of the head motion. Despite its helpfulness in general, this traditional scrubbing scheme cannot guarantee that the removed volumes are necessarily unhelpful, since such a step is fully independent to the subsequent BFN estimation task. Moreover, the removal of volumes would reduce the statistical power, and different numbers of volumes are generally scrubbed for different subjects, resulting in an inconsistency or bias in the estimated BFNs. To address these issues, we develop a new learning framework that conducts BFN estimation and data-scrubbing simultaneously by an alternating optimization algorithm. The newly developed algorithm adaptively weights volumes (instead of removing them directly) for the task of BFN estimation. As a result, the proposed method can not only reduce the difficulty of threshold selection involved in the traditional scrubbing scheme, but also provide a more flexible framework that scrubs the data in the subsequent FBN estimation model. Finally, we validate the proposed method by identifying subjects with mild cognitive impairment (MCI) from normal controls based on the estimated BFNs, achieving an 80.22% classification accuracy, which significantly improves the baseline methods.
Background: Functional brain networks (FBNs) derived from resting-state functional MRI (rs-fMRI) have shown great potential in identifying brain disorders, such as autistic spectrum disorder (ASD). Therefore, many FBN estimation methods have been proposed in recent years. Most existing methods only model the functional connections between brain regions of interest (ROIs) from a single view (e.g., by estimating FBNs through a specific strategy), failing to capture the complex interactions among ROIs in the brain. Methods: To address this problem, we propose fusion of multiview FBNs through joint embedding, which can make full use of the common information of multiview FBNs estimated by different strategies. More specifically, we first stack the adjacency matrices of FBNs estimated by different methods into a tensor and use tensor factorization to learn the joint embedding (i.e., a common factor of all FBNs) for each ROI. Then, we use Pearson’s correlation to calculate the connections between each embedded ROI in order to reconstruct a new FBN. Results: Experimental results obtained on the public ABIDE dataset with rs-fMRI data reveal that our method is superior to several state-of-the-art methods in automated ASD diagnosis. Moreover, by exploring FBN “features” that contributed most to ASD identification, we discovered potential biomarkers for ASD diagnosis. The proposed framework achieves an accuracy of , which is generally better than the compared individual FBN methods. In addition, our method achieves the best performance compared to other multinetwork methods, i.e., an accuracy improvement of at least . Conclusions: We present a multiview FBN fusion strategy through joint embedding for fMRI-based ASD identification. The proposed fusion method has an elegant theoretical explanation from the perspective of eigenvector centrality.