Synthetic CT (sCT) generated from CBCT has proven effective in artifact reduction and CT number correction, facilitating precise radiation dose calculation. However, the quality of different regions in sCT images is severely imbalanced, with soft tissue region exhibiting notably inferior quality compared to others. To address this imbalance, we proposed a Multi-Task Attention Network (MuTA-Net) based on VGG-16, specifically focusing the enhancement of image quality in soft tissue region of sCT. First, we introduced a multi-task learning strategy that divides the sCT generation task into three sub-tasks: global image generation, soft tissue region generation and bone region segmentation. This approach ensured the quality of overall sCT image while enhancing the network's focus on feature extraction and generation for soft tissues region. The result of bone region segmentation task guided the fusion of sub-tasks results. Then, we designed an attention module to further optimize feature extraction capabilities of the network. Finally, by employing a results fusion module, the results of three sub-tasks were integrated, generating a high-quality sCT image. Experimental results on head and neck CBCT demonstrated that the sCT images generated by the proposed MuTA-Net exhibited a 12.52% reduction in mean absolute error in soft tissue region, compared to the best performance among the three comparative methods, including ResNet, U-Net, and U-Net++. It can be seen that MuTA-Net is suitable for high-quality sCT image generation and has potential application value in the field of CBCT guided adaptive radiation therapy.
Cone-Beam Computed Tomography/methods* ; Humans ; Image Processing, Computer-Assisted/methods* ; Artifacts ; Algorithms ; Bone and Bones/diagnostic imaging* ; Neural Networks, Computer

Alzheimer's disease (AD) classification models usually segment the entire brain image into voxel blocks and assign them labels consistent with the entire image, but not every voxel block is closely related to the disease. To this end, an AD auxiliary diagnosis framework based on weakly supervised multi-instance learning (MIL) and multi-scale feature fusion is proposed, and the framework is designed from three aspects: within the voxel block, between voxel blocks, and high-confidence voxel blocks. First, a three-dimensional convolutional neural network was used to extract deep features within the voxel block; then the spatial correlation information between voxel blocks was captured through position encoding and attention mechanism; finally, high-confidence voxel blocks were selected and combined with multi-scale information fusion strategy to integrate key features for classification decision. The performance of the model was evaluated on the Alzheimer's Disease Neuroimaging Initiative (ADNI) and Open Access Series of Imaging Studies (OASIS) datasets. Experimental results showed that the proposed framework improved ACC and AUC by 3% and 4% on average compared with other mainstream frameworks in the two tasks of AD classification and mild cognitive impairment conversion classification, and could find the key voxel blocks that trigger the disease, providing an effective basis for AD auxiliary diagnosis.
Alzheimer Disease/diagnosis* ; Humans ; Neuroimaging/methods* ; Neural Networks, Computer ; Brain/diagnostic imaging* ; Magnetic Resonance Imaging ; Deep Learning ; Machine Learning

Due to its irregular shape and varying contour, pancreas segmentation is a recognized challenge in medical image segmentation. Convolutional neural network (CNN) and Transformer-based networks perform well but have limitations: CNN have constrained receptive fields, and Transformer underutilize image features. This work proposes an improved pancreas segmentation method by combining CNN and Transformer. Point-wise separable convolution was introduced in a stage-wise encoder to extract more features with fewer parameters. A densely connected ensemble decoder enabled multi-scale feature fusion, addressing the structural constraints of skip connections. Consistency terms and contrastive loss were integrated into deep supervision to ensure model accuracy. Extensive experiments on the Changhai and National Institute of Health (NIH) pancreas datasets achieved the highest Dice similarity coefficient (DSC) values of 76.32% and 86.78%, with superiority in other metrics. Ablation studies validated each component's contributions to performance and parameter reduction. Results demonstrate that the proposed loss function smooths training and optimizes performance. Overall, the method outperforms other advanced methods, enhances pancreas segmentation performance, supports physician diagnosis, and provides a reliable reference for future research.
Humans ; Neural Networks, Computer ; Pancreas/diagnostic imaging* ; Image Processing, Computer-Assisted/methods* ; Algorithms ; Deep Learning

The accurate segmentation of breast ultrasound images is an important precondition for the lesion determination. The existing segmentation approaches embrace massive parameters, sluggish inference speed, and huge memory consumption. To tackle this problem, we propose T ²KD Attention U-Net (dual-Teacher Knowledge Distillation Attention U-Net), a lightweight semantic segmentation method combined double-path joint distillation in breast ultrasound images. Primarily, we designed two teacher models to learn the fine-grained features from each class of images according to different feature representation and semantic information of benign and malignant breast lesions. Then we leveraged the joint distillation to train a lightweight student model. Finally, we constructed a novel weight balance loss to focus on the semantic feature of small objection, solving the unbalance problem of tumor and background. Specifically, the extensive experiments conducted on Dataset BUSI and Dataset B demonstrated that the T ²KD Attention U-Net outperformed various knowledge distillation counterparts. Concretely, the accuracy, recall, precision, Dice, and mIoU of proposed method were 95.26%, 86.23%, 85.09%, 83.59%and 77.78% on Dataset BUSI, respectively. And these performance indexes were 97.95%, 92.80%, 88.33%, 88.40% and 82.42% on Dataset B, respectively. Compared with other models, the performance of this model was significantly improved. Meanwhile, compared with the teacher model, the number, size, and complexity of student model were significantly reduced (2.2×10 ⁶ vs. 106.1×10 ⁶, 8.4 MB vs. 414 MB, 16.59 GFLOPs vs. 205.98 GFLOPs, respectively). Indeedy, the proposed model guarantees the performances while greatly decreasing the amount of computation, which provides a new method for the deployment of clinical medical scenarios.
Humans ; Breast Neoplasms/diagnostic imaging* ; Female ; Ultrasonography, Mammary/methods* ; Image Processing, Computer-Assisted/methods* ; Algorithms ; Neural Networks, Computer ; Breast/diagnostic imaging*

Real-time acquisition of pulmonary ventilation and perfusion information through thoracic electrical impedance tomography (EIT) holds significant clinical value. This study proposes a novel method based on the rime (RIME) algorithm-optimized variational mode decomposition (VMD) to separate lung ventilation and perfusion signals directly from raw voltage data prior to EIT image reconstruction, enabling independent imaging of both parameters. To validate this approach, EIT data were collected from 16 healthy volunteers under normal breathing and inspiratory breath-holding conditions. The RIME algorithm was employed to optimize VMD parameters by minimizing envelope entropy as the fitness function. The optimized VMD was then applied to separate raw data across all measurement channels in EIT, with spectral analysis identifying relevant components to reconstruct ventilation and perfusion signals. Results demonstrated that the structural similarity index (SSIM) between perfusion images derived from normal breathing and breath-holding states averaged approximately 84% across all 16 subjects, significantly outperforming traditional frequency-domain filtering methods in perfusion imaging accuracy. This method offers a promising technical advancement for real-time monitoring of pulmonary ventilation and perfusion, holding significant value for advancing the clinical application of EIT in the diagnosis and treatment of respiratory diseases.
Humans ; Electric Impedance ; Algorithms ; Tomography/methods* ; Pulmonary Ventilation/physiology* ; Lung/diagnostic imaging* ; Image Processing, Computer-Assisted/methods* ; Adult

Aiming at the problems of low accuracy and large difference of segmentation boundary distance in anterior cruciate ligament (ACL) image segmentation of knee joint, this paper proposes an ACL image segmentation model by fusing dilated convolution and residual hybrid attention U-shaped network (DRH-UNet). The proposed model builds upon the U-shaped network (U-Net) by incorporating dilated convolutions to expand the receptive field, enabling a better understanding of the contextual relationships within the image. Additionally, a residual hybrid attention block is designed in the skip connections to enhance the expression of critical features in key regions and reduce the semantic gap, thereby improving the representation capability for the ACL area. This study constructs an enhanced annotated ACL dataset based on the publicly available Magnetic Resonance Imaging Network (MRNet) dataset. The proposed method is validated on this dataset, and the experimental results demonstrate that the DRH-UNet model achieves a Dice similarity coefficient (DSC) of (88.01±1.57)% and a Hausdorff distance (HD) of 5.16±0.85, outperforming other ACL segmentation methods. The proposed approach further enhances the segmentation accuracy of ACL, providing valuable assistance for subsequent clinical diagnosis by physicians.
Humans ; Magnetic Resonance Imaging/methods* ; Anterior Cruciate Ligament/diagnostic imaging* ; Image Processing, Computer-Assisted/methods* ; Knee Joint/diagnostic imaging* ; Neural Networks, Computer ; Algorithms ; Deep Learning

Among numerous medical imaging modalities, diffusion weighted imaging (DWI) is extremely sensitive to acute ischemic stroke lesions, especially small infarcts. However, magnetic resonance imaging is time-consuming and expensive, and it is also prone to interference from metal implants. Therefore, the aim of this study is to design a medical image synthesis method based on generative adversarial network, Stroke-p2pHD, for synthesizing DWI images from computed tomography (CT). Stroke-p2pHD consisted of a generator that effectively fused local image features and global context information (Global_to_Local) and a multi-scale discriminator (M ²Dis). Specifically, in the Global_to_Local generator, a fully convolutional Transformer (FCT) and a local attention module (LAM) were integrated to achieve the synthesis of detailed information such as textures and lesions in DWI images. In the M ²Dis discriminator, a multi-scale convolutional network was adopted to perform the discrimination function of the input images. Meanwhile, an optimization balance with the Global_to_Local generator was ensured and the consistency of features in each layer of the M ²Dis discriminator was constrained. In this study, the public Acute Ischemic Stroke Dataset (AISD) and the acute cerebral infarction dataset from Yantaishan Hospital were used to verify the performance of the Stroke-p2pHD model in synthesizing DWI based on CT. Compared with other methods, the Stroke-p2pHD model showed excellent quantitative results (mean-square error = 0.008, peak signal-to-noise ratio = 23.766, structural similarity = 0.743). At the same time, relevant experimental analyses such as computational efficiency verify that the Stroke-p2pHD model has great potential for clinical applications.
Humans ; Tomography, X-Ray Computed/methods* ; Diffusion Magnetic Resonance Imaging/methods* ; Cerebral Infarction/diagnostic imaging* ; Stroke/diagnostic imaging* ; Neural Networks, Computer ; Image Processing, Computer-Assisted/methods* ; Algorithms

Retrieving keyframes most relevant to text from small intestine videos with given labels can efficiently and accurately locate pathological regions. However, training directly on raw video data is extremely slow, while learning visual representations from image-text datasets leads to computational inconsistency. To tackle this challenge, a small bowel video keyframe retrieval based on multi-modal contrastive learning (KRCL) is proposed. This framework fully utilizes textual information from video category labels to learn video features closely related to text, while modeling temporal information within a pretrained image-text model. It transfers knowledge learned from image-text multimodal models to the video domain, enabling interaction among medical videos, images, and text data. Experimental results on the hyper-spectral and Kvasir dataset for gastrointestinal disease detection (Hyper-Kvasir) and the Microsoft Research video-to-text (MSR-VTT) retrieval dataset demonstrate the effectiveness and robustness of KRCL, with the proposed method achieving state-of-the-art performance across nearly all evaluation metrics.
Humans ; Video Recording ; Intestine, Small/diagnostic imaging* ; Machine Learning ; Image Processing, Computer-Assisted/methods* ; Algorithms

Online hashing methods are receiving increasing attention in cross modal medical image retrieval research. However, existing online methods often lack the learning ability to maintain semantic correlation between new and existing data. To this end, we proposed online semantic similarity cross-modal hashing (OSCMH) learning framework to incrementally learn compact binary hash codes of medical stream data. Within it, a sparse representation of existing data based on online anchor datasets was designed to avoid semantic forgetting of the data and adaptively update hash codes, which effectively maintained semantic correlation between existing and arriving data and reduced information loss as well as improved training efficiency. Besides, an online discrete optimization method was proposed to solve the binary optimization problem of hash code by incrementally updating hash function and optimizing hash code on medical stream data. Compared with existing online or offline hashing methods, the proposed algorithm achieved average retrieval accuracy improvements of 12.5% and 14.3% on two datasets, respectively, effectively enhancing the retrieval efficiency in the field of medical images.
Semantics ; Humans ; Algorithms ; Information Storage and Retrieval/methods* ; Diagnostic Imaging ; Image Processing, Computer-Assisted/methods*

To address the high computational complexity of the Transformer in the segmentation of ultrasound thyroid nodules and the loss of image details or omission of key spatial information caused by traditional image sampling techniques when dealing with high-resolution, complex texture or uneven density two-dimensional ultrasound images, this paper proposes a thyroid nodule segmentation method that integrates the receiving weighted key-value (RWKV) architecture and spherical geometry feature (SGF) sampling technology. This method effectively captures the details of adjacent regions through two-dimensional offset prediction and pixel-level sampling position adjustment, achieving precise segmentation. Additionally, this study introduces a patch attention module (PAM) to optimize the decoder feature map using a regional cross-attention mechanism, enabling it to focus more precisely on the high-resolution features of the encoder. Experiments on the thyroid nodule segmentation dataset (TN3K) and the digital database for thyroid images (DDTI) show that the proposed method achieves dice similarity coefficients (DSC) of 87.24% and 80.79% respectively, outperforming existing models while maintaining a lower computational complexity. This approach may provide an efficient solution for the precise segmentation of thyroid nodules.
Thyroid Nodule/diagnostic imaging* ; Humans ; Ultrasonography/methods* ; Algorithms ; Image Processing, Computer-Assisted/methods* ; Thyroid Gland/diagnostic imaging*