Multi-hop question answering requires retrieving and integrating evidence from multiple contexts.Despite the rapid progress of current research, multi-hop reasoning remains constrained by two persistent limitations: attention decay, where the model's focus on main question degrades as the reasoning chain grows, and error accumulation, where mistakes propagate across hops and compounds into final failure.Inspired by Theta-Gamma hierarchical oscillation which decouples global planning from local retrieval, enabling efficient attention transfer between hops and a verification and repair mechanism that interrupts the accumulation of errors in the wrong paths, we present THOR, a brain-inspired Theta-Gamma hierarchical oscillatory reasoning framework.Extensive comparative experiments and specific validation experiments on multi-hop QA benchmarks demonstrate that THOR improves answer accuracy and robustness while mitigating limitations, showcasing its generalization across different backbones.

Dataset distillation (DD) aims to improve training efficiency by condensing large datasets into compact yet informative subsets. Existing DD methods primarily use optimization-based approaches for image synthesis, which are computationally intensive. While some recent studies have explored optimization-free alternatives, their simplistic region selection strategies result in poor representations of the original dataset and inefficient utilization of synthetic data during downstream training. To address these limitations, we propose a method called Selection-and-Guidance for Dataset Distillation (SGDD). This approach refines the distillation process through two key stages: region selection and guidance enhancement. Specifically, we first obtain a candidate set of regions from various locations within the images. Then, we utilize the Representative Region Selector and Diverse Region Selector to identify the critical regions for image classification. Furthermore, we generate multiview guidance information for the synthetic data to enhance the distillation process further. By selecting representative and diverse regions while incorporating multiview guidance, our method unleashes the potential of optimization-free DD. Experimental results substantiate the superiority of our approach across various datasets and network architectures.

ABSTRACT Low‐quality fundus images pose significant challenges for both ophthalmologists and computer‐aided diagnosis systems. While many existing deep learning‐based image quality enhancement algorithms require low‐ and high‐quality image pairs for training, such pairs are often difficult to obtain in practice. On the other hand, unpaired image enhancement algorithms tend to struggle in preserving small structures and suppressing artefacts, which are crucial for medical applications. To address these issues, we propose an unpaired structure‐preserving cycle quality alternating network for low‐quality fundus image enhancement. Our method consists of three main components: (1) a cycle quality alternating framework to provide pixel‐wise supervision for unpaired image enhancement, (2) a quality‐aware disentangle module to enhance the extrinsic representation of the low‐quality image with the high‐quality reference image, and (3) an instance normalized skip to improve the network's structure‐preserving capability. We tested our method on both synthetic and authentic clinical images with pathological structures and found it to be superior to state‐of‐the‐art algorithms in terms of improving image quality while preserving delicate structures. Additionally, the proposed network demonstrated strong generalization ability in improving the quality of unseen images, as tested on 135‐degree neonatal fundus images.

Cervical abnormal cell detection plays a crucial role in the early screening of cervical cancer. In recent years, some deep learning-based methods have been proposed. However, these methods rely heavily on large amounts of annotated images, which are time-consuming and labor-intensive to acquire, thus limiting the detection performance. In this paper, we present a novel Semi-supervised Cervical Abnormal Cell detector (SCAC), which effectively utilizes the abundant unlabeled data. We utilize Transformer as the backbone of SCAC to capture long-range dependencies to mimic the diagnostic process of pathologists. In addition, in SCAC, we design a Unified Strong and Weak Augment strategy (USWA) that unifies two data augmentation pipelines, implementing consistent regularization in semi-supervised learning and enhancing the diversity of the training data. We also develop a Global Attention Feature Pyramid Network (GAFPN), which utilizes the attention mechanism to better extract multi-scale features from cervical cytology images. Notably, we have created an unlabeled cervical cytology image dataset, which can be leveraged by semi-supervised learning to enhance detection accuracy. To the best of our knowledge, this is the first publicly available large unlabeled cervical cytology image dataset. By combining this dataset with two publicly available annotated datasets, we demonstrate that SCAC outperforms other existing methods, achieving state-of-the-art performance. Additionally, comprehensive ablation studies are conducted to validate the effectiveness of USWA and GAFPN. These promising results highlight the capability of SCAC to achieve high diagnostic accuracy and extensive clinical applications.

Retinal arteriovenous nicking (AVN) manifests as a reduced venular caliber of an arteriovenous crossing. AVNs are signs of many systemic, particularly cardiovascular diseases. Studies have shown that people with AVN are twice as likely to have a stroke. However, AVN classification faces two challenges. One is the lack of data, especially AVNs compared to the normal arteriovenous (AV) crossings. The other is the significant intra-class variations and minute inter-class differences. AVNs may look different in shape, scale, pose, and color. On the other hand, the AVN could be different from the normal AV crossing only by slight thinning of the vein. To address these challenges, first, we develop a data synthesis method to generate AV crossings, including normal and AVNs. Second, to mitigate the domain shift between the synthetic and real data, an edge-guided unsupervised domain adaptation network is designed to guide the transfer of domain invariant information. Third, a semantic contrastive learning branch (SCLB) is introduced and a set of semantically related images, as a semantic triplet, are input to the network simultaneously to guide the network to focus on the subtle differences in venular width and to ignore the differences in appearance. These strategies effectively mitigate the lack of data, domain shift between synthetic and real data, and significant intra- but minute inter-class differences. Extensive experiments have been performed to demonstrate the outstanding performance of the proposed method.

Deep convolutional neural network (DCNN) models have been widely explored for skin disease diagnosis and some of them have achieved the diagnostic outcomes comparable or even superior to those of dermatologists. However, broad implementation of DCNN in skin disease detection is hindered by small size and data imbalance of the publically accessible skin lesion datasets. This paper proposes a novel single-model based strategy for classification of skin lesions on small and imbalanced datasets. First, various DCNNs are trained on different small and imbalanced datasets to verify that the models with moderate complexity outperform the larger models. Second, regularization DropOut and DropBlock are added to reduce overfitting and a Modified RandAugment augmentation strategy is proposed to deal with the defects of sample underrepresentation in the small dataset. Finally, a novel Multi-Weighted New Loss (MWNL) function and an end-to-end cumulative learning strategy (CLS) are introduced to overcome the challenge of uneven sample size and classification difficulty and to reduce the impact of abnormal samples on training. By combining Modified RandAugment, MWNL and CLS, our single DCNN model method achieved the classification accuracy comparable or superior to those of multiple ensembling models on different dermoscopic image datasets. Our study shows that this method is able to achieve a high classification performance at a low cost of computational resources and inference time, potentially suitable to implement in mobile devices for automated screening of skin lesions and many other malignancies in low resource settings.