Lately, deep learning has been extensively investigated for accelerating dynamic magnetic resonance (MR) imaging, with encouraging progresses achieved. However, without fully sampled reference data for training, current approaches may have limited abilities in recovering fine details or structures. To address this challenge, this paper proposes a self-supervised collaborative learning framework (SelfCoLearn) for accurate dynamic MR image reconstruction from undersampled k-space data. The proposed framework is equipped with three important components, namely, dual-network collaborative learning, reunderampling data augmentation and a specially designed co-training loss. The framework is flexible to be integrated with both data-driven networks and model-based iterative un-rolled networks. Our method has been evaluated on in-vivo dataset and compared it to four state-of-the-art methods. Results show that our method possesses strong capabilities in capturing essential and inherent representations for direct reconstructions from the undersampled k-space data and thus enables high-quality and fast dynamic MR imaging.
Graph Neural Networks (GNNs) is a promising approach for applications with nonEuclidean data. However, training GNNs on large scale graphs with hundreds of millions nodes is both resource and time consuming. Different from DNNs, GNNs usually have larger memory footprints, and thus the GPU memory capacity and PCIe bandwidth are the main resource bottlenecks in GNN training. To address this problem, we present BiFeat: a graph feature quantization methodology to accelerate GNN training by significantly reducing the memory footprint and PCIe bandwidth requirement so that GNNs can take full advantage of GPU computing capabilities. Our key insight is that unlike DNN, GNN is less prone to the information loss of input features caused by quantization. We identify the main accuracy impact factors in graph feature quantization and theoretically prove that BiFeat training converges to a network where the loss is within $\epsilon$ of the optimal loss of uncompressed network. We perform extensive evaluation of BiFeat using several popular GNN models and datasets, including GraphSAGE on MAG240M, the largest public graph dataset. The results demonstrate that BiFeat achieves a compression ratio of more than 30 and improves GNN training speed by 200%-320% with marginal accuracy loss. In particular, BiFeat achieves a record by training GraphSAGE on MAG240M within one hour using only four GPUs.
Vision Transformers have achieved state-of-the-art performance in many visual tasks. Due to the quadratic computational and memory complexities of self-attention, recent works either apply attention only to low-resolution inputs or restrict the receptive field to a small local region. To overcome these limitations, we propose key-only attention, which excludes query-key pairwise interactions and uses a compute-efficient saliency-gate to obtain attention weights, modeling local-global interactions in all stages. Key-only attention has linear computational and memory complexities w.r.t input size. We use alternate layout to hybridize convolution and attention layers instead of grafting which is suggested by previous works, so that all stages can benefit from both spatial attentions and convolutions. We leverage these improvements to develop a new self-attention model family, LinGlos, which reach state-of-the-art accuracies on the parameter-limited setting of ImageNet classification benchmark, and outperform baselines significantly in downstream tasks, e.g., COCO object detection and ADE20K semantic segmentation.
The past several years have witnessed the success of transformer-based models, and their scale and application scenarios continue to grow aggressively. The current landscape of transformer models is increasingly diverse: the model size varies drastically with the largest being of hundred-billion parameters; the model characteristics differ due to the sparsity introduced by the Mixture-of-Experts; the target application scenarios can be latency-critical or throughput-oriented; the deployment hardware could be single- or multi-GPU systems with different types of memory and storage, etc. With such increasing diversity and the fast-evolving pace of transformer models, designing a highly performant and efficient inference system is extremely challenging. In this paper, we present DeepSpeed Inference, a comprehensive system solution for transformer model inference to address the above-mentioned challenges. DeepSpeed Inference consists of (1) a multi-GPU inference solution to minimize latency while maximizing the throughput of both dense and sparse transformer models when they fit in aggregate GPU memory, and (2) a heterogeneous inference solution that leverages CPU and NVMe memory in addition to the GPU memory and compute to enable high inference throughput with large models which do not fit in aggregate GPU memory. DeepSpeed Inference reduces latency by up to 7.3X over the state-of-the-art for latency-oriented scenarios and increases throughput by over 1.5x for throughput-oriented scenarios. Moreover, it enables trillion parameter scale inference under real-time latency constraints by leveraging hundreds of GPUs, an unprecedented scale for inference. It can inference 25x larger models than with GPU-only solutions, while delivering a high throughput of 84 TFLOPS (over $50\%$ of A6000 peak).
Ghosting artifacts, motion blur, and low fidelity in highlight are the main challenges in High Dynamic Range (HDR) imaging from multiple Low Dynamic Range (LDR) images. These issues come from using the medium-exposed image as the reference frame in previous methods. To deal with them, we propose to use the under-exposed image as the reference to avoid these issues. However, the heavy noise in dark regions of the under-exposed image becomes a new problem. Therefore, we propose a joint HDR and denoising pipeline, containing two sub-networks: (i) a pre-denoising network (PreDNNet) to adaptively denoise input LDRs by exploiting exposure priors; (ii) a pyramid cascading fusion network (PCFNet), introducing an attention mechanism and cascading structure in a multi-scale manner. To further leverage these two paradigms, we propose a selective and joint HDR and denoising (SJ-HD$^2$R) imaging framework, utilizing scenario-specific priors to conduct the path selection with an accuracy of more than 93.3$\%$. We create the first joint HDR and denoising benchmark dataset, which contains a variety of challenging HDR and denoising scenes and supports the switching of the reference image. Extensive experiment results show that our method achieves superior performance to previous methods.
Recovering high-quality images from undersampled measurements is critical for accelerated MRI reconstruction. Recently, various supervised deep learning-based MRI reconstruction methods have been developed. Despite the achieved promising performances, these methods require fully sampled reference data, the acquisition of which is resource-intensive and time-consuming. Self-supervised learning has emerged as a promising solution to alleviate the reliance on fully sampled datasets. However, existing self-supervised methods suffer from reconstruction errors due to the insufficient constraint enforced on the non-sampled data points and the error accumulation happened alongside the iterative image reconstruction process for model-driven deep learning reconstrutions. To address these challenges, we propose K2Calibrate, a K-space adaptation strategy for self-supervised model-driven MR reconstruction optimization. By iteratively calibrating the learned measurements, K2Calibrate can reduce the network's reconstruction deterioration caused by statistically dependent noise. Extensive experiments have been conducted on the open-source dataset FastMRI, and K2Calibrate achieves better results than five state-of-the-art methods. The proposed K2Calibrate is plug-and-play and can be easily integrated with different model-driven deep learning reconstruction methods.
With the successful application of deep learning in magnetic resonance imaging, parallel imaging techniques based on neural networks have attracted wide attentions. However, without high-quality fully sampled datasets for training, the performance of these methods tends to be limited. To address this issue, this paper proposes a physics based unsupervised contrastive representation learning (PARCEL) method to speed up parallel MR imaging. Specifically, PARCEL has three key ingredients to achieve direct deep learning from the undersampled k-space data. Namely, a parallel framework has been developed by learning two branches of model-based networks unrolled with the conjugate gradient algorithm; Augmented undersampled k-space data randomly drawn from the obtained k-space data are used to help the parallel network to capture the detailed information. A specially designed co-training loss is designed to guide the two networks to capture the inherent features and representations of the-to-be-reconstructed MR image. The proposed method has been evaluated on in vivo datasets and compared to five state-of-the-art methods, whose results show PARCEL is able to learn useful representations for more accurate MR reconstructions without the reliance on the fully-sampled datasets.
Radiomics and deep learning have shown high popularity in automatic glioma grading. Radiomics can extract hand-crafted features that quantitatively describe the expert knowledge of glioma grades, and deep learning is powerful in extracting a large number of high-throughput features that facilitate the final classification. However, the performance of existing methods can still be improved as their complementary strengths have not been sufficiently investigated and integrated. Furthermore, lesion maps are usually needed for the final prediction at the testing phase, which is very troublesome. In this paper, we propose an expert knowledge-guided geometric representation learning (ENROL) framework . Geometric manifolds of hand-crafted features and learned features are constructed to mine the implicit relationship between deep learning and radiomics, and therefore to dig mutual consent and essential representation for the glioma grades. With a specially designed manifold discrepancy measurement, the grading model can exploit the input image data and expert knowledge more effectively in the training phase and get rid of the requirement of lesion segmentation maps at the testing phase. The proposed framework is flexible regarding deep learning architectures to be utilized. Three different architectures have been evaluated and five models have been compared, which show that our framework can always generate promising results.
Prostate Imaging Reporting and Data System (PI-RADS) based on multi-parametric MRI classi\^ees patients into 5 categories (PI-RADS 1-5) for routine clinical diagnosis guidance. However, there is no consensus on whether PI-RADS 3 patients should go through biopsies. Mining features from these hard samples (HS) is meaningful for physicians to achieve accurate diagnoses. Currently, the mining of HS biomarkers is insu\`icient, and the e\'eectiveness and robustness of HS biomarkers for prostate cancer diagnosis have not been explored. In this study, biomarkers from di\'eerent data distributions are constructed. Results show that HS biomarkers can achieve better performances in di\'eerent data distributions.
Radiation therapy (RT) is widely employed in the clinic for the treatment of head and neck (HaN) cancers. An essential step of RT planning is the accurate segmentation of various organs-at-risks (OARs) in HaN CT images. Nevertheless, segmenting OARs manually is time-consuming, tedious, and error-prone considering that typical HaN CT images contain tens to hundreds of slices. Automated segmentation algorithms are urgently required. Recently, convolutional neural networks (CNNs) have been extensively investigated on this task. Particularly, 3D CNNs are frequently adopted to process 3D HaN CT images. There are two issues with na\"ive 3D CNNs. First, the depth resolution of 3D CT images is usually several times lower than the in-plane resolution. Direct employment of 3D CNNs without distinguishing this difference can lead to the extraction of distorted image features and influence the final segmentation performance. Second, a severe class imbalance problem exists, and large organs can be orders of times larger than small organs. It is difficult to simultaneously achieve accurate segmentation for all the organs. To address these issues, we propose a novel hybrid CNN that fuses 2D and 3D convolutions to combat the different spatial resolutions and extract effective edge and semantic features from 3D HaN CT images. To accommodate large and small organs, our final model, named OrganNet2.5D, consists of only two instead of the classic four downsampling operations, and hybrid dilated convolutions are introduced to maintain the respective field. Experiments on the MICCAI 2015 challenge dataset demonstrate that OrganNet2.5D achieves promising performance compared to state-of-the-art methods.