Accurate segmentation and classification of colorectal polyps are critical for the early diagnosis and prevention of colorectal cancer. However, high variability in polyp appearance, along with object-background and inter-class visual similarities, presents significant challenges for automated analysis. In this paper, we propose SC-UMamba, a unified deep learning architecture that integrates a CNN backbone with a Mamba-based UNet-like architecture to jointly address segmentation and classification tasks. Specifically, SC-UMamba leverages shared feature representations between the SSM encoder and CNN backbone, and further incorporates semantic guidance from the segmentation mask output. Additionally, pretrained weights and a step-based curriculum learning strategy are adopted to enable robust learning across both tasks. Furthermore, the architecture offers modular flexibility to adapt to different clinical requirements by supporting backbone substitution. Extensive experiments on the SUN-SEG and our proprietary clinical dataset demonstrate that SC-UMamba achieves state-of-the-art performance, outperforming existing methods in both segmentation accuracy and classification robustness. These results highlight the potential of SC-UMamba to enhance real-time, intelligent assistance during colonoscopy procedures and support reliable clinical decision-making.

Medical anomaly detection is challenged by the scarcity and diversity of abnormal samples and the lack of precise annotations, limiting the scalability of supervised methods. To address this, we propose FMA-GEN, a novel diffusion-based framework for few-shot, multi-conditional controllable medical anomaly generation and detection. Specifically, FMA-GEN first introduces a text-guided anomaly mask generator, allowing the creation of diverse and semantically meaningful masks. These masks, combined with anomaly embeddings and textual prompts, are used to condition the diffusion process, facilitating the synthesis of high-fidelity and controllable anomalous images. To ensure anatomical realism, we further design a boundary-aware blending module that fuses normal and abnormal regions along mask boundaries. Finally, in the downstream detection stage, we develop a semi-supervised learning scheme that leverages the generated samples to enhance anomaly representation. A mask-guided feature mining strategy is employed to highlight discriminative abnormal features while suppressing interference from normal regions. Extensive experiments on the BraTS and LiverCT datasets demonstrate that FMA-GEN generates realistic and diverse anomalies, leading to significant improvements in both anomaly detection and localization tasks. Our code are available at https://github.com/TytopiaAI/FMA-GEN.

Colorectal cancer (CRC) is a prevalent malignancy with significant social and healthcare implications, ranking among the top causes of cancer-related mortality worldwide. In this computational social systems context, we address the challenge of accurately segmenting CRC lesions from colonoscopy images, which is pivotal for early cancer detection and treatment. The complexity of intestinal environments and variations in medical expertise contribute to the high rate of undetected or misdiagnosed lesions, underscoring the need for advanced computational models. This study presents ColoSegNet, a novel self-supervised deep learning framework designed to enhance the accuracy and effectiveness of CRC diagnosis. By leveraging a comprehensive and annotated colorectal lesion segmentation dataset (CLSD), ColoSegNet incorporates a temporal correlation module to extract critical features from colonoscopy video frames, significantly improving the segmentation of colorectal lesions. Furthermore, ColoSegNet employs a masked autoencoder (MAE) module for self-supervised image reconstruction, preserving the original image integrity and facilitating precise segmentation. Comparative assessments against established models such as UNet, PraNet, and Deeplab V3 demonstrate ColoSegNet’s superior performance in detailed feature representation and overall segmentation accuracy. This research not only contributes to the field of medical imaging but also to computational social systems, by capturing inherent data patterns and integrating specialized modules for feature representation in a healthcare context. Our findings provide valuable insights into the early and accurate detection of CRC, a critical issue given the disease’s high incidence and mortality rates, and its impact on social systems.

Precise segmentation of the pupil, iris, and sclera is critical for diagnosing and treating ocular diseases such as glaucoma, strabismus, and retinal disorders. However, the fine structural differences within the eye and the interference of complex backgrounds, especially with VR devices prone to reflections, tilts, distortions, and occlusions, present significant challenges. In this paper, we introduce SS-SwinUnet, a novel segmentation method that integrates Swin Transformer and knowledge distillation to achieve superior performance. Specifically, SS-SwinUnet balances feature transfer between the encoder and decoder, reducing redundancy and enhancing representation. Additionally, we incorporate a Boundary Difference over Union Loss to improve boundary segmentation accuracy. We also propose an eye modeling method that parameterizes segmentation results to optimize the semantic segmentation of ocular structures. We constructed the TongRenD dataset, comprising 400 VR-captured videos and 4,100 images, which, along with the TEyeD dataset, was used in our experiments. Results demonstrate that SS-SwinUnet significantly outperforms existing medical image segmentation methods across multiple datasets.

Sublingual vein features including swelling, varicose and cyanosis are essential for the symptoms differentiation and treatment selection in Traditional Chinese Medicine (TCM) tongue diagnosis, especially reflecting the state of human blood circulation. However, automatic and accurate extraction of sublingual vein features remains a great challenge, limited by both the lack of datasets for sublingual images and the influence of noise from non-tongue and non-sublingual vein components. In this paper, we propose a novel tongue features extraction method based on segmenting the sublingual vein instead of the whole tongue bottom, in which a sublingual vein segmentation framework based on a Polyp-PVT network is developed to eliminate the noise from the surrounding part of the sublingual vein. Meanwhile, we first adopt a transformer-based method such as Swin-Transformer network to extract sublingual vein features by virtue of the awesome capability of the transformer network. In addition, we construct a large dataset including 4018 sublingual vein images for the segmentation and classification of sublingual veins. Experimental results have shown that the tongue feature extraction method combined with a sublingual vein segmentation can greatly outperform the existing tongue feature extracting methods.

Checkpointing is a key technique to guarantee execution correctness and ensure progress forwarding in energy harvesting systems. However, checkpointing itself introduces system overhead due to extra operations of data movements between volatile memory and nonvolatile memory. Moreover, execution rollback to the latest checkpoint under a power failure can cancel some obtained progress and this waste is highly correlated to the latest checkpoint interval. These two kinds of overhead can be quite high if the checkpoint interval setting mismatches the power input characteristic. Unlike previous checkpointing schemes emphasizing on optimizing data copy overhead, this paper further takes into account the characteristic of input power sources and proposes a resilient checkpointing scheme, ResCheck, which can capture the power input changes and thus accordingly mitigate checkpoint overhead and rollback cost. The proposed ResCheck, directed by a lightweight neural network-based power level predictor, is capable of adjusting the checkpoint intervals to fit different power levels at runtime. In this way, an energy harvesting system equipped with ResCheck can achieve both fewer checkpoint number and lower execution rollback punishment. Our experimental results show that ResCheck can reduce an average checkpoint number of 24.4% and 14.6% over the conventional periodic checkpointing scheme and the state-of-the-art iCheck scheme respectively. Meanwhile, ResCheck improves average performance as well as energy efficiency by more than three times compared to iCheck.

With the growth of the convolutional neural network (CNN) parameters, the hardware resources become limited when deploying CNN models. Single bit-width quantization may lead to degradation of accuracy, while mixed-precision quantization models maintain higher accuracy. However, current mixed-precision quantization accelerators only consider the algorithm level quantization and fixed-bit-width processing elements (PEs) without fully utilizing the resources. Therefore, we propose a neural network acceleration architecture based on bit-level computational units to improve resource utilization and throughput of mixed-precision quantization accelerators. The 2-bit and 3-bit low-bit computational units are designed to implement the high-bit quantization. We also propose spatio-temporal fusion to satisfy the unique bit widths in each layer with mixed precision quantization. In particular, we use the 2-bit and 3-bit computational units to achieve dynamic layer-level quantization. We also discuss various combinations of different bit widths, which can be dynamically implemented according to the requirements of accuracy, execution time, etc. The proposed acceleration architecture is implemented in Verilog, verified using three network models: ResNet18, VGG7 and LeNet-5, and tested against three accelerators: Eyeriss, Stripes and Bit Fusion. The experimental results show that our accelerators provide more accuracy and the grouping operation reduces the area overhead. It provides a 3.08 to 3.54 times acceleration ratio and 2.4 to 2.57 times reduction in energy consumption on the three networks.

Multiplying matrices is among the most fundamental and compute-intensive operations in machine learning. Approximated Matrix Multiplication (AMM) based on table look-ups can significantly reduce the pressure on computing units and memory bandwidth, and has great potential in large-scale machine learning applications. In this work, we speed up table look-ups on GPUs to improve the performance of matrix multiplication. To avoid random memory accesses in table look-ups, we propose a novel warp-wide data sharing execution model. With this execution model, we develop a GPU AMM library to speed up MADDNESS (the state-of-the-art AMM), named GPU-MADDNESS. The experimental results show that GPU-MADDNESS improves the performance by 103X on average, and outperforms the tiling implementation by up to 42%.

The amount of impervious surface is an important indicator to measure the degree of urbanization and the urban ecological environment. However, the objects in the low-density impervious surface areas are small and scattered, which are easily confused with the background. Therefore, the extraction of the small and scattered impervious surfaces is still challenging. In this study, we propose a dual-branch network combing transformer and CNN with attention mechanism. In this model, transformer branch is first used to extract impervious surface to capture long-distance and large-scale dependencies. In addition, another UNet branch embedded the coordinate attention mechanism can capture detailed information and meanwhile reduce information redundancy. Experiments show that our proposed method performs better than the traditional CNN methods.

Accurate polyp segmentation from colonoscopy images is essential for identifying colorectal cancer. Recently, segmentation methods based on convolutional neural networks and transformers have represented excellent performance for image polyp segmentation. However, these methods are mostly designed for individual images rather than the entire video datasets, which results in the absence of sequential relationships among lesion images and neglects the significant intrinsic property of continuous video. In this work, we propose a temporal correlation network (TC-Net) for video polyp segmentation. In TC-Net, the temporal correlation is unprecedentedly modeled based on the relationship between the original video and the captured frames to be adaptable for video polyp segmentation, and the network is also calibrated for the corresponding time correlation output. Furthermore, we design a dual-track learning strategy for the optimization method in TC-Net to ensure the independence of TC-Net during the learning process to adequately exploit the optimization effect of temporal correlation. The network’s effectiveness is demonstrated by extensive experiments on five publicly available biomedical datasets, and TC-Net achieves state-of-the-art (SOTA) performance.

Impervious surface extraction from remote sensing images has become a promising technology to measure the urban ecological environment and monitor human activity. However, due to the complex characteristics of impervious landscapes, most researches on impervious surface extraction hardly identify the scattered and small objects especially in difficult areas, which severely affect the accuracy of mapping impervious surface. In this work, we propose a channel and spatial attention mechanism (CSAM) to extract impervious surface in difficult areas, which includes a channel attention module to learn the relationship in the multi-channel remote sensing images and a spatial attention module to capture the features of the inconspicuous objects. Experiments with the Sentinel-2 dataset in South Africa demonstrate that CSAM can outperform the state-of-the-art methods.

Tongue diagnosis plays an essential role in diagnosing the syndrome types, pathological types, lesion location and clinical stages of cancers in Traditional Chinese Medicine (TCM). The quality of the tongue image datasets is crucial to tongue image segmentation in modern tongue diagnosis. However, the tongue image dataset maintains a challenging problem because of the lack of datasets for the sublingual image, the complexity and scarcity of the tongue images. In this paper, we propose an effective segmentation framework for tongue images, called Moment Invariants with Data Augmentation (DAMI), which can be flexibly applied to different segmentation models. To overcome over-fitting caused by excessive data augmentation, a data augmentation module with random transformations is designed to achieve appropriate data augmentation. Meanwhile, we develop a new moment invariants module to optimize data augmentation in image segmentation. In addition, a novel tongue image dataset, Lingual-Sublingual Image Dataset (LSID), has been established for the classification and segmentation of tongue or sublingual veins. Experimental results confirm that the models using DAMI can remarkably outperform the existing methods on LID, SID and BioHit for the tongue image segmentation.

Cloud data centers provide end-users with a wide range of application scenarios, including scientific computing, smart grids, etc. The number and size of data centers have rapidly increased in recent years, which causes severe environmental problems and colossal power demand. Therefore, it is desirable to use a proper scheduling method to optimize resource usage and reduce energy consumption in a data center. However, it is rather difficult to design an effective and efficient task scheduling algorithm because of the dynamic and complex environment of data centers. This paper proposes a task scheduling algorithm, WSS, to optimize resource usage and reduce energy consumption based on a model-free deep reinforcement learning framework inspired by the Wolpertinger architecture. The proposed algorithm can handle the scheduling problem on a sizeable discrete action space, improve decision efficiency, and save the training convergence time. Meanwhile, the proposed algorithm based on Soft Actor-Critic is designed to improve the stability and exploration capability of WSS. Experiments based on real-world traces prove that WSS can reduce energy consumption by nearly 25% compared with the Deep Q-network task scheduling algorithm. Moreover, WSS can provide a short time of training convergence without increasing the average waiting time of tasks and achieve stable performance.