Datasets for biosignals, such as electroencephalogram (EEG) and electrocardiogram (ECG), often have noisy labels and have limited number of subjects (<100). To handle these challenges, we propose a self-supervised approach based on contrastive learni
ng to model biosignals with a reduced reliance on labeled data and with fewer subjects. In this regime of limited labels and subjects, intersubject variability negatively impacts model performance. Thus, we introduce subject-aware learning through (1) a subject-specific contrastive loss, and (2) an adversarial training to promote subject-invariance during the self-supervised learning. We also develop a number of time-series data augmentation techniques to be used with the contrastive loss for biosignals. Our method is evaluated on publicly available datasets of two different biosignals with different tasks: EEG decoding and ECG anomaly detection. The embeddings learned using self-supervision yield competitive classification results compared to entirely supervised methods. We show that subject-invariance improves representation quality for these tasks, and observe that subject-specific loss increases performance when fine-tuning with supervised labels.
Many real-world signal sources are complex-valued, having real and imaginary components. However, the vast majority of existing deep learning platforms and network architectures do not support the use of complex-valued data. MRI data is inherently co
mplex-valued, so existing approaches discard the richer algebraic structure of the complex data. In this work, we investigate end-to-end complex-valued convolutional neural networks - specifically, for image reconstruction in lieu of two-channel real-valued networks. We apply this to magnetic resonance imaging reconstruction for the purpose of accelerating scan times and determine the performance of various promising complex-valued activation functions. We find that complex-valued CNNs with complex-valued convolutions provide superior reconstructions compared to real-valued convolutions with the same number of trainable parameters, over a variety of network architectures and datasets.
Purpose: To study the accuracy of motion information extracted from beat-to-beat 3D image-based navigators (3D iNAVs) collected using a variable-density cones trajectory with different combinations of spatial resolutions and scan acceleration factors
. Methods: Fully sampled, breath-held 4.4 mm 3D iNAV datasets for six respiratory phases are acquired in a volunteer. Ground truth translational and nonrigid motion information is derived from these datasets. Subsequently, the motion estimates from synthesized undersampled 3D iNAVs with isotropic spatial resolutions of 4.4 mm (acceleration factor = 10.9), 5.4 mm (acceleration factor = 7.2), 6.4 mm (acceleration factor = 4.2), and 7.8 mm (acceleration factor = 2.9) are assessed against the ground truth information. The undersampled 3D iNAV configuration with the highest accuracy motion estimates in simulation is then compared with the originally proposed 4.4 mm undersampled 3D iNAV in six volunteer studies. Results: The simulations indicate that for navigators beyond certain scan acceleration factors, the accuracy of motion estimates is compromised due to errors from residual aliasing and blurring/smoothening effects following compressed sensing reconstruction. The 6.4 mm 3D iNAV achieves an acceptable spatial resolution with a small acceleration factor, resulting in the highest accuracy motion information among all assessed undersampled 3D iNAVs. Reader scores for six volunteer studies demonstrate superior coronary vessel sharpness when applying an autofocusing nonrigid correction technique using the 6.4 mm 3D iNAVs in place of 4.4 mm 3D iNAVs. Conclusion: Undersampled 6.4 mm 3D iNAVs enable motion tracking with improved accuracy relative to previously proposed undersampled 4.4 mm 3D iNAVs.
Purpose: To develop a framework to reconstruct large-scale volumetric dynamic MRI from rapid continuous and non-gated acquisitions, with applications to pulmonary and dynamic contrast enhanced (DCE) imaging. Theory and Methods: The problem consider
ed here requires recovering hundred-gigabytes of dynamic volumetric image data from a few gigabytes of k-space data, acquired continuously over several minutes. This reconstruction is vastly under-determined, heavily stressing computing resources as well as memory management and storage. To overcome these challenges, we leverage intrinsic three dimensional (3D) trajectories, such as 3D radial and 3D cones, with ordering that incoherently cover time and k-space over the entire acquisition. We then propose two innovations: (1) A compressed representation using multi-scale low rank matrix factorization that constrains the reconstruction problem, and reduces its memory footprint. (2) Stochastic optimization to reduce computation, improve memory locality, and minimize communications between threads and processors. We demonstrate the feasibility of the proposed method on DCE imaging acquired with a golden-angle ordered 3D cones trajectory and pulmonary imaging acquired with a bit-reversed ordered 3D radial trajectory. We compare it with soft-gated dynamic reconstruction for DCE and respiratory resolved reconstruction for pulmonary imaging. Results: The proposed technique shows transient dynamics that are not seen in gating based methods. When applied to datasets with irregular, or non-repetitive motions, the proposed method displays sharper image features. Conclusion: We demonstrated a method that can reconstruct massive 3D dynamic image series in the extreme undersampling and extreme computation setting.
Magnetic resonance image (MRI) reconstruction is a severely ill-posed linear inverse task demanding time and resource intensive computations that can substantially trade off {it accuracy} for {it speed} in real-time imaging. In addition, state-of-the
-art compressed sensing (CS) analytics are not cognizant of the image {it diagnostic quality}. To cope with these challenges we put forth a novel CS framework that permeates benefits from generative adversarial networks (GAN) to train a (low-dimensional) manifold of diagnostic-quality MR images from historical patients. Leveraging a mixture of least-squares (LS) GANs and pixel-wise $ell_1$ cost, a deep residual network with skip connections is trained as the generator that learns to remove the {it aliasing} artifacts by projecting onto the manifold. LSGAN learns the texture details, while $ell_1$ controls the high-frequency noise. A multilayer convolutional neural network is then jointly trained based on diagnostic quality images to discriminate the projection quality. The test phase performs feed-forward propagation over the generator network that demands a very low computational overhead. Extensive evaluations are performed on a large contrast-enhanced MR dataset of pediatric patients. In particular, images rated based on expert radiologists corroborate that GANCS retrieves high contrast images with detailed texture relative to conventional CS, and pixel-wise schemes. In addition, it offers reconstruction under a few milliseconds, two orders of magnitude faster than state-of-the-art CS-MRI schemes.
