Deep learning has potential to automate screening, monitoring and grading of disease in medical images. Pretraining with contrastive learning enables models to extract robust and generalisable features from natural image datasets, facilitating label-efficient downstream image analysis. However, the direct application of conventional contrastive methods to medical datasets introduces two domain-specific issues. Firstly, several image transformations which have been shown to be crucial for effective contrastive learning do not translate from the natural image to the medical image domain. Secondly, the assumption made by conventional methods, that any two images are dissimilar, is systematically misleading in medical datasets depicting the same anatomy and disease. This is exacerbated in longitudinal image datasets that repeatedly image the same patient cohort to monitor their disease progression over time. In this paper we tackle these issues by extending conventional contrastive frameworks with a novel metadata-enhanced strategy. Our approach employs widely available patient metadata to approximate the true set of inter-image contrastive relationships. To this end we employ records for patient identity, eye position (i.e. left or right) and time series information. In experiments using two large longitudinal datasets containing 170,427 retinal optical coherence tomography (OCT) images of 7912 patients with age-related macular degeneration (AMD), we evaluate the utility of using metadata to incorporate the temporal dynamics of disease progression into pretraining. Our metadata-enhanced approach outperforms both standard contrastive methods and a retinal image foundation model in five out of six image-level downstream tasks related to AMD. We find benefits in both a low-data and high-data regime across tasks ranging from AMD stage and type classification to prediction of visual acuity. Due to its modularity, our method can be quickly and cost-effectively tested to establish the potential benefits of including available metadata in contrastive pretraining.

The burgeoning field of brain health research increasingly leverages artificial intelligence (AI) to analyze and interpret neuroimaging data. Medical foundation models have shown promise of superior performance with better sample efficiency. This work introduces a novel approach towards creating 3-dimensional (3D) medical foundation models for multimodal neuroimage segmentation through self-supervised training. Our approach involves a novel two-stage pretraining approach using vision transformers. The first stage encodes anatomical structures in generally healthy brains from the large-scale unlabeled neuroimage dataset of multimodal brain magnetic resonance imaging (MRI) images from 41,400 participants. This stage of pertaining focuses on identifying key features such as shapes and sizes of different brain structures. The second pretraining stage identifies disease-specific attributes, such as geometric shapes of tumors and lesions and spatial placements within the brain. This dual-phase methodology significantly reduces the extensive data requirements usually necessary for AI model training in neuroimage segmentation with the flexibility to adapt to various imaging modalities. We rigorously evaluate our model, BrainSegFounder, using the Brain Tumor Segmentation (BraTS) challenge and Anatomical Tracings of Lesions After Stroke v2.0 (ATLAS v2.0) datasets. BrainSegFounder demonstrates a significant performance gain, surpassing the achievements of the previous winning solutions using fully supervised learning. Our findings underscore the impact of scaling up both the model complexity and the volume of unlabeled training data derived from generally healthy brains. Both of these factors enhance the accuracy and predictive capabilities of the model in neuroimage segmentation tasks. Our pretrained models and code are at https://github.com/lab-smile/BrainSegFounder.

The rapid dissemination of unverified information through social platforms like Twitter poses considerable dangers to societal stability. Identifying real versus fake claims is challenging, and previous work on rumor detection methods often fails to effectively capture propagation structure features. These methods also often overlook the presence of comments irrelevant to the discussion topic of the source post. To address this, we introduce a novel approach: the Structure-Aware Multilevel Graph Attention Network (SAMGAT) for rumor classification. SAMGAT employs a dynamic attention mechanism that blends GATv2 and dot-product attention to capture the contextual relationships between posts, allowing for varying attention scores based on the stance of the central node. The model incorporates a structure-aware attention mechanism that learns attention weights that can indicate the existence of edges, effectively reflecting the propagation structure of rumors. Moreover, SAMGAT incorporates a top-k attention filtering mechanism to select the most relevant neighboring nodes, enhancing its ability to focus on the key structural features of rumor propagation. Furthermore, SAMGAT includes a claim-guided attention pooling mechanism with a thresholding step to focus on the most informative posts when constructing the event representation. Experimental results on benchmark datasets demonstrate that SAMGAT outperforms state-of-the-art methods in identifying rumors and improves the effectiveness of early rumor detection.

Traumatic Brain Injury (TBI) presents a broad spectrum of clinical presentations and outcomes due to its inherent heterogeneity, leading to diverse recovery trajectories and varied therapeutic responses. While many studies have delved into TBI phenotyping for distinct patient populations, identifying TBI phenotypes that consistently generalize across various settings and populations remains a critical research gap. Our research addresses this by employing multivariate time-series clustering to unveil TBI\'s dynamic intricates. Utilizing a self-supervised learning-based approach to clustering multivariate time-Series data with missing values (SLAC-Time), we analyzed both the research-centric TRACK-TBI and the real-world MIMIC-IV datasets. Remarkably, the optimal hyperparameters of SLAC-Time and the ideal number of clusters remained consistent across these datasets, underscoring SLAC-Time\'s stability across heterogeneous datasets. Our analysis revealed three generalizable TBI phenotypes (α, β, and γ), each exhibiting distinct non-temporal features during emergency department visits, and temporal feature profiles throughout ICU stays. Specifically, phenotype α represents mild TBI with a remarkably consistent clinical presentation. In contrast, phenotype β signifies severe TBI with diverse clinical manifestations, and phenotype γ represents a moderate TBI profile in terms of severity and clinical diversity. Age is a significant determinant of TBI outcomes, with older cohorts recording higher mortality rates. Importantly, while certain features varied by age, the core characteristics of TBI manifestations tied to each phenotype remain consistent across diverse populations.

Pediatric respiratory disease diagnosis and subsequent treatment require accurate and interpretable analysis. A chest X-ray is the most cost-effective and rapid method for identifying and monitoring various thoracic diseases in children. Recent developments in self-supervised and transfer learning have shown their potential in medical imaging, including chest X-ray areas. In this article, we propose a three-stage framework with knowledge transfer from adult chest X-rays to aid the diagnosis and interpretation of pediatric thorax diseases. We conducted comprehensive experiments with different pre-training and fine-tuning strategies to develop transformer or convolutional neural network models and then evaluate them qualitatively and quantitatively. The ViT-Base/16 model, fine-tuned with the CheXpert dataset, a large chest X-ray dataset, emerged as the most effective, achieving a mean AUC of 0.761 (95% CI: 0.759-0.763) across six disease categories and demonstrating a high sensitivity (average 0.639) and specificity (average 0.683), which are indicative of its strong discriminative ability. The baseline models, ViT-Small/16 and ViT-Base/16, when directly trained on the Pediatric CXR dataset, only achieved mean AUC scores of 0.646 (95% CI: 0.641-0.651) and 0.654 (95% CI: 0.648-0.660), respectively. Qualitatively, our model excels in localizing diseased regions, outperforming models pre-trained on ImageNet and other fine-tuning approaches, thus providing superior explanations. The source code is available online and the data can be obtained from PhysioNet.

Nowadays, autonomous driving technology has become widely prevalent. The intelligent vehicles have been equipped with various sensors (e.g., vision sensors, LiDAR, depth cameras etc.). Among them, the vision systems with tailored semantic segmentation and perception algorithms play critical roles in scene understanding. However, the traditional supervised semantic segmentation needs a large number of pixel-level manual annotations to complete model training. Although few-shot methods reduce the annotation work to some extent, they are still labor intensive. In this paper, a self-supervised few-shot semantic segmentation method based on Multi-task Learning and Dense Attention Computation (dubbed MLDAC) is proposed. The salient part of an image is split into two parts; one of them serves as the support mask for few-shot segmentation, while cross-entropy losses are calculated between the other part and the entire region with the predicted results separately as multi-task learning so as to improve the model\'s generalization ability. Swin Transformer is used as our backbone to extract feature maps at different scales. These feature maps are then input to multiple levels of dense attention computation blocks to enhance pixel-level correspondence. The final prediction results are obtained through inter-scale mixing and feature skip connection. The experimental results indicate that MLDAC obtains 55.1% and 26.8% one-shot mIoU self-supervised few-shot segmentation on the PASCAL-5i and COCO-20i datasets, respectively. In addition, it achieves 78.1% on the FSS-1000 few-shot dataset, proving its efficacy.

Robust object detection in complex environments, poor visual conditions, and open scenarios presents significant technical challenges in autonomous driving. These challenges necessitate the development of advanced fusion methods for millimeter-wave (mmWave) radar point cloud data and visual images. To address these issues, this paper proposes a radar-camera robust fusion network (RCRFNet), which leverages self-supervised learning and open-set recognition to effectively utilise the complementary information from both sensors. Specifically, the network uses matched radar-camera data through a frustum association approach to generate self-supervised signals, enhancing network training. The integration of global and local depth consistencies between radar point clouds and visual images, along with image features, helps construct object class confidence levels for detecting unknown targets. Additionally, these techniques are combined with a multi-layer feature extraction backbone and a multimodal feature detection head to achieve robust object detection. Experiments on the nuScenes public dataset demonstrate that RCRFNet outperforms state-of-the-art (SOTA) methods, particularly in conditions of low visual visibility and when detecting unknown class objects.

The implementation of deep learning in Magnetic Resonance Imaging (MRI) has significantly advanced the reduction of data acquisition times. However, these techniques face substantial limitations in scenarios where acquiring fully sampled datasets is unfeasible or costly. To tackle this problem, we propose a Fusion enhanced Contrastive Self-Supervised Learning (FCSSL) method for parallel MRI reconstruction, eliminating the need for fully sampled k-space training dataset and coil sensitivity maps. First, we introduce a strategy based on two pairs of re-undersampling masks within a contrastive learning framework, aimed at enhancing the representational capacity to achieve higher quality reconstruction. Subsequently, a novel adaptive fusion network, trained in a self-supervised learning manner, is designed to integrate the reconstruction results of the framework. Experimental results on knee datasets under different sampling masks demonstrate that the proposed FCSSL achieves superior reconstruction performance compared to other self-supervised learning methods. Moreover, the performance of FCSSL approaches that of the supervised methods, especially under the 2DRU and RADU masks. The proposed FCSSL, trained under the 3× 1DRU and 2DRU masks, can effectively generalize to unseen 1D and 2D undersampling masks, respectively. For target domain data that exhibit significant differences from source domain data, the proposed model, fine-tuned with just a few dozen instances of undersampled data in the target domain, achieves reconstruction performance comparable to that achieved by the model trained with the entire set of undersampled data. The novel FCSSL model offers a viable solution for reconstructing high-quality MR images without needing fully sampled datasets, thereby overcoming a major hurdle in scenarios where acquiring fully sampled MR data is difficult.

Sequential recommendation typically utilizes deep neural networks to mine rich information in interaction sequences. However, existing methods often face the issue of insufficient interaction data. To alleviate the sparsity issue, self-supervised learning is introduced into sequential recommendation. Despite its effectiveness, we argue that current self-supervised learning-based (i.e., SSL-based) sequential recommendation models have the following limitations: (1) using only a single self-supervised learning method, either contrastive self-supervised learning or generative self-supervised learning. (2) employing a simple data augmentation strategy in either the graph structure domain or the node feature domain. We believe that they have not fully utilized the capabilities of both self-supervised methods and have not sufficiently explored the advantages of combining graph augmentation schemes. As a result, they often fail to learn better item representations. In light of this, we propose a novel multi-task sequential recommendation framework named Adaptive Self-supervised Learning for sequential Recommendation (ASLRec). Specifically, our framework combines contrastive and generative self-supervised learning methods adaptively, simultaneously applying different perturbations at both the graph topology and node feature levels. This approach constructs diverse augmented graph views and employs multiple loss functions (including contrastive loss, generative loss, mask loss, and prediction loss) for joint training. By encompassing the capabilities of various methods, our model learns item representations across different augmented graph views to achieve better performance and effectively mitigate interaction noise and sparsity. In addition, we add a small proportion of random uniform noise to item representations, making the item representations more uniform and mitigating the inherent popularity bias in interaction records. We conduct extensive experiments on three publicly available benchmark datasets to evaluate our model. The results demonstrate that our approach achieves state-of-the-art performance compared to 14 other competitive methods: the hit rate (HR) improved by over 14.39%, and the normalized discounted cumulative gain (NDCG) increased by over 18.67%.

Shape registration of patient-specific organ shapes to endoscopic camera images is expected to be a key to realizing image-guided surgery, and a variety of applications of machine learning methods have been considered. Because the number of training data available from clinical cases is limited, the use of synthetic images generated from a statistical deformation model has been attempted; however, the influence on estimation caused by the difference between synthetic images and real scenes is a problem. In this study, we propose a self-supervised offline learning framework for model-based registration using image features commonly obtained from synthetic images and real camera images. Because of the limited number of endoscopic images available for training, we use a synthetic image generated from the nonlinear deformation model that represents possible intraoperative pneumothorax deformations. In order to solve the difficulty in estimating deformed shapes and viewpoints from the common image features obtained from synthetic and real images, we attempted to improve the registration error by adding the shading and distance information that can be obtained as prior knowledge in the synthetic image. Shape registration with real camera images is performed by learning the task of predicting the differential model parameters between two synthetic images. The developed framework achieved registration accuracy with a mean absolute error of less than 10 mm and a mean distance of less than 5 mm in a thoracoscopic pulmonary cancer resection, confirming improved prediction accuracy compared with conventional methods.