Attribute Surrogates Learning and Spectral Tokens Pooling in Transformers for Few-shot Learning

This paper presents new hierarchically cascaded transformers that can improve data efficiency through attribute surrogates learning and spectral tokens pooling. Vision transformers have recently been thought of as a promising alternative to convolutional neural networks for visual recognition. But when there is no sufficient data, it gets stuck in overfitting and shows inferior performance. To improve data efficiency, we propose hierarchically cascaded transformers that exploit intrinsic image structures through spectral tokens pooling and optimize the learnable parameters through latent attribute surrogates. The intrinsic image structure is utilized to reduce the ambiguity between foreground content and background noise by spectral tokens pooling. And the attribute surrogate learning scheme is designed to benefit from the rich visual information in image-label pairs instead of simple visual concepts assigned by their labels. Our Hierarchically Cascaded Transformers, called HCTransformers, is built upon a self-supervised learning framework DINO and is tested on several popular few-shot learning benchmarks. In the inductive setting, HCTransformers surpass the DINO baseline by a large margin of 9.7% 5-way 1-shot accuracy and 9.17% 5-way 5-shot accuracy on miniImageNet, which demonstrates HCTransformers are efficient to extract discriminative features. Also, HCTransformers show clear advantages over SOTA few-shot classification methods in both 5-way 1-shot and 5-way 5-shot settings on four popular benchmark datasets, including miniImageNet, tieredImageNet, FC100, and CIFAR-FS. The trained weights and codes are available at https://github.com/StomachCold/HCTransformers.

PointMatch: A Consistency Training Framework for Weakly SupervisedSemantic Segmentation of 3D Point Clouds

Semantic segmentation of point cloud usually relies on dense annotation that is exhausting and costly, so it attracts wide attention to investigate solutions for the weakly supervised scheme with only sparse points annotated. Existing works start from the given labels and propagate them to highly-related but unlabeled points, with the guidance of data, e.g. intra-point relation. However, it suffers from (i) the inefficient exploitation of data information, and (ii) the strong reliance on labels thus is easily suppressed when given much fewer annotations. Therefore, we propose a novel framework, PointMatch, that stands on both data and label, by applying consistency regularization to sufficiently probe information from data itself and leveraging weak labels as assistance at the same time. By doing so, meaningful information can be learned from both data and label for better representation learning, which also enables the model more robust to the extent of label sparsity. Simple yet effective, the proposed PointMatch achieves the state-of-the-art performance under various weakly-supervised schemes on both ScanNet-v2 and S3DIS datasets, especially on the settings with extremely sparse labels, e.g. surpassing SQN by 21.2% and 17.2% on the 0.01% and 0.1% setting of ScanNet-v2, respectively.

BOAT: Bilateral Local Attention Vision Transformer

Vision Transformers achieved outstanding performance in many computer vision tasks. Early Vision Transformers such as ViT and DeiT adopt global self-attention, which is computationally expensive when the number of patches is large. To improve efficiency, recent Vision Transformers adopt local self-attention mechanisms, where self-attention is computed within local windows. Despite the fact that window-based local self-attention significantly boosts efficiency, it fails to capture the relationships between distant but similar patches in the image plane. To overcome this limitation of image-space local attention, in this paper, we further exploit the locality of patches in the feature space. We group the patches into multiple clusters using their features, and self-attention is computed within every cluster. Such feature-space local attention effectively captures the connections between patches across different local windows but still relevant. We propose a Bilateral lOcal Attention vision Transformer (BOAT), which integrates feature-space local attention with image-space local attention. We further integrate BOAT with both Swin and CSWin models, and extensive experiments on several benchmark datasets demonstrate that our BOAT-CSWin model clearly and consistently outperforms existing state-of-the-art CNN models and vision Transformers.

Real-time automatic polyp detection in colonoscopy using feature enhancement module and spatiotemporal similarity correlation unit

Automatic detection of polyps is challenging because different polyps vary greatly, while the changes between polyps and their analogues are small. The state-of-the-art methods are based on convolutional neural networks (CNNs). However, they may fail due to lack of training data, resulting in high rates of missed detection and false positives (FPs). In order to solve these problems, our method combines the two-dimensional (2-D) CNN-based real-time object detector network with spatiotemporal information. Firstly, we use a 2-D detector network to detect static images and frames, and based on the detector network, we propose two feature enhancement modules-the FP Relearning Module (FPRM) to make the detector network learning more about the features of FPs for higher precision, and the Image Style Transfer Module (ISTM) to enhance the features of polyps for sensitivity improvement. In video detection, we integrate spatiotemporal information, which uses Structural Similarity (SSIM) to measure the similarity between video frames. Finally, we propose the Inter-frame Similarity Correlation Unit (ISCU) to combine the results obtained by the detector network and frame similarity to make the final decision. We verify our method on both private databases and publicly available databases. Experimental results show that these modules and units provide a performance improvement compared with the baseline method. Comparison with the state-of-the-art methods shows that the proposed method outperforms the existing ones which can meet real-time constraints. It's demonstrated that our method provides a performance improvement in sensitivity, precision and specificity, and has great potential to be applied in clinical colonoscopy.

Cross-Modality Deep Feature Learning for Brain Tumor Segmentation

Recent advances in machine learning and prevalence of digital medical images have opened up an opportunity to address the challenging brain tumor segmentation (BTS) task by using deep convolutional neural networks. However, different from the RGB image data that are very widespread, the medical image data used in brain tumor segmentation are relatively scarce in terms of the data scale but contain the richer information in terms of the modality property. To this end, this paper proposes a novel cross-modality deep feature learning framework to segment brain tumors from the multi-modality MRI data. The core idea is to mine rich patterns across the multi-modality data to make up for the insufficient data scale. The proposed cross-modality deep feature learning framework consists of two learning processes: the cross-modality feature transition (CMFT) process and the cross-modality feature fusion (CMFF) process, which aims at learning rich feature representations by transiting knowledge across different modality data and fusing knowledge from different modality data, respectively. Comprehensive experiments are conducted on the BraTS benchmarks, which show that the proposed cross-modality deep feature learning framework can effectively improve the brain tumor segmentation performance when compared with the baseline methods and state-of-the-art methods.

Advancing 3D Medical Image Analysis with Variable Dimension Transform based Supervised 3D Pre-training

The difficulties in both data acquisition and annotation substantially restrict the sample sizes of training datasets for 3D medical imaging applications. As a result, constructing high-performance 3D convolutional neural networks from scratch remains a difficult task in the absence of a sufficient pre-training parameter. Previous efforts on 3D pre-training have frequently relied on self-supervised approaches, which use either predictive or contrastive learning on unlabeled data to build invariant 3D representations. However, because of the unavailability of large-scale supervision information, obtaining semantically invariant and discriminative representations from these learning frameworks remains problematic. In this paper, we revisit an innovative yet simple fully-supervised 3D network pre-training framework to take advantage of semantic supervisions from large-scale 2D natural image datasets. With a redesigned 3D network architecture, reformulated natural images are used to address the problem of data scarcity and develop powerful 3D representations. Comprehensive experiments on four benchmark datasets demonstrate that the proposed pre-trained models can effectively accelerate convergence while also improving accuracy for a variety of 3D medical imaging tasks such as classification, segmentation and detection. In addition, as compared to training from scratch, it can save up to 60% of annotation efforts. On the NIH DeepLesion dataset, it likewise achieves state-of-the-art detection performance, outperforming earlier self-supervised and fully-supervised pre-training approaches, as well as methods that do training from scratch. To facilitate further development of 3D medical models, our code and pre-trained model weights are publicly available at https://github.com/urmagicsmine/CSPR.

Generalized Radiograph Representation Learning via Cross-supervision between Images and Free-text Radiology Reports

Pre-training lays the foundation for recent successes in radiograph analysis supported by deep learning. It learns transferable image representations by conducting large-scale fully-supervised or self-supervised learning on a source domain. However, supervised pre-training requires a complex and labor intensive two-stage human-assisted annotation process while self-supervised learning cannot compete with the supervised paradigm. To tackle these issues, we propose a cross-supervised methodology named REviewing FreE-text Reports for Supervision (REFERS), which acquires free supervision signals from original radiology reports accompanying the radiographs. The proposed approach employs a vision transformer and is designed to learn joint representations from multiple views within every patient study. REFERS outperforms its transfer learning and self-supervised learning counterparts on 4 well-known X-ray datasets under extremely limited supervision. Moreover, REFERS even surpasses methods based on a source domain of radiographs with human-assisted structured labels. Thus REFERS has the potential to replace canonical pre-training methodologies.

Symmetry-Enhanced Attention Network for Acute Ischemic Infarct Segmentation with Non-Contrast CT Images

Quantitative estimation of the acute ischemic infarct is crucial to improve neurological outcomes of the patients with stroke symptoms. Since the density of lesions is subtle and can be confounded by normal physiologic changes, anatomical asymmetry provides useful information to differentiate the ischemic and healthy brain tissue. In this paper, we propose a symmetry enhanced attention network (SEAN) for acute ischemic infarct segmentation. Our proposed network automatically transforms an input CT image into the standard space where the brain tissue is bilaterally symmetric. The transformed image is further processed by a Ushape network integrated with the proposed symmetry enhanced attention for pixel-wise labelling. The symmetry enhanced attention can efficiently capture context information from the opposite side of the image by estimating long-range dependencies. Experimental results show that the proposed SEAN outperforms some symmetry-based state-of-the-art methods in terms of both dice coefficient and infarct localization.

Context-LGM: Leveraging Object-Context Relation for Context-Aware Object Recognition

Context, as referred to situational factors related to the object of interest, can help infer the object's states or properties in visual recognition. As such contextual features are too diverse (across instances) to be annotated, existing attempts simply exploit image labels as supervision to learn them, resulting in various contextual tricks, such as features pyramid, context attention, etc. However, without carefully modeling the context's properties, especially its relation to the object, their estimated context can suffer from large inaccuracy. To amend this problem, we propose a novel Contextual Latent Generative Model (Context-LGM), which considers the object-context relation and models it in a hierarchical manner. Specifically, we firstly introduce a latent generative model with a pair of correlated latent variables to respectively model the object and context, and embed their correlation via the generative process. Then, to infer contextual features, we reformulate the objective function of Variational Auto-Encoder (VAE), where contextual features are learned as a posterior distribution conditioned on the object. Finally, to implement this contextual posterior, we introduce a Transformer that takes the object's information as a reference and locates correlated contextual factors. The effectiveness of our method is verified by state-of-the-art performance on two context-aware object recognition tasks, i.e. lung cancer prediction and emotion recognition.

nnFormer: Interleaved Transformer for Volumetric Segmentation

Transformers, the default model of choices in natural language processing, have drawn scant attention from the medical imaging community. Given the ability to exploit long-term dependencies, transformers are promising to help atypical convolutional neural networks (convnets) to overcome its inherent shortcomings of spatial inductive bias. However, most of recently proposed transformer-based segmentation approaches simply treated transformers as assisted modules to help encode global context into convolutional representations without investigating how to optimally combine self-attention (i.e., the core of transformers) with convolution. To address this issue, in this paper, we introduce nnFormer (i.e., Not-aNother transFormer), a powerful segmentation model with an interleaved architecture based on empirical combination of self-attention and convolution. In practice, nnFormer learns volumetric representations from 3D local volumes. Compared to the naive voxel-level self-attention implementation, such volume-based operations help to reduce the computational complexity by approximate 98% and 99.5% on Synapse and ACDC datasets, respectively. In comparison to prior-art network configurations, nnFormer achieves tremendous improvements over previous transformer-based methods on two commonly used datasets Synapse and ACDC. For instance, nnFormer outperforms Swin-UNet by over 7 percents on Synapse. Even when compared to nnUNet, currently the best performing fully-convolutional medical segmentation network, nnFormer still provides slightly better performance on Synapse and ACDC.