Dual-TranSpeckle: Dual-pathway transformer based encoder-decoder network for medical ultrasound image despeckling.

The majority of existing deep learning-based image denoising algorithms mainly focus on processing the overall image features, ignoring the fine differences between the semantic and pixel features. Hence, we propose Dual-TranSpeckle (DTS), a medical ultrasound image despeckling network built on a dual-path Transformer. The DTS introduces two different paths, named "semantic path" and "pixel path," to facilitate the parallel transfer of feature information within the image. The semantic path passes a global view of the input semantic features, and the image features are passed through a Semantic Block to extract global semantic information from pixel-level features. The pixel path is employed to transmit finer-grained pixel features. Within the dual-path network framework, two essential modules, namely Dual Block and Merge Block, are designed. These leverage the Transformer architecture during the encoding and decoding stages. The Dual Block module facilitates information interaction between the semantic and pixel features by considering the interdependencies across both paths. Meanwhile, the Merge Block module enables parallel transfer of feature information by merging the dual path features, thereby facilitating the self-attention calculations for the overall feature representation. Our DTS is extensively evaluated on two public datasets and one private dataset. The DTS network demonstrates significant enhancements in quantitative evaluation results in terms of peak signal-to-noise ratio (PSNR), structural similarity (SSIM), feature similarity (FSIM), and naturalness image quality evaluator (NIQE). Furthermore, our qualitative analysis confirms that the DTS has significant improvements in despeckling performance, effectively suppressing speckle noise while preserving essential image structures.

HCformer: Hybrid CNN-Transformer for LDCT Image Denoising.

Low-dose computed tomography (LDCT) is an effective way to reduce radiation exposure for patients. However, it will increase the noise of reconstructed CT images and affect the precision of clinical diagnosis. The majority of the current deep learning-based denoising methods are built on convolutional neural networks (CNNs), which concentrate on local information and have little capacity for multiple structures modeling. Transformer structures are capable of computing each pixel's response on a global scale, but their extensive computation requirements prevent them from being widely used in medical image processing. To reduce the impact of LDCT scans on patients, this paper aims to develop an image post-processing method by combining CNN and Transformer structures. This method can obtain a high-quality images from LDCT. A hybrid CNN-Transformer (HCformer) codec network model is proposed for LDCT image denoising. A neighborhood feature enhancement (NEF) module is designed to introduce the local information into the Transformer's operation, and the representation of adjacent pixel information in the LDCT image denoising task is increased. The shifting window method is utilized to lower the computational complexity of the network model and overcome the problems that come with computing the MSA (Multi-head self-attention) process in a fixed window. Meanwhile, W/SW-MSA (Windows/Shifted window Multi-head self-attention) is alternately used in two layers of the Transformer to gain the information interaction between various Transformer layers. This approach can successfully decrease the Transformer's overall computational cost. The AAPM 2016 LDCT grand challenge dataset is employed for ablation and comparison experiments to demonstrate the viability of the proposed LDCT denoising method. Per the experimental findings, HCformer can increase the image quality metrics SSIM, HuRMSE and FSIM from 0.8017, 34.1898, and 0.6885 to 0.8507, 17.7213, and 0.7247, respectively. Additionally, the proposed HCformer algorithm will preserves image details while it reduces noise. In this paper, an HCformer structure is proposed based on deep learning and evaluated by using the AAPM LDCT dataset. Both the qualitative and quantitative comparison results confirm that the proposed HCformer outperforms other methods. The contribution of each component of the HCformer is also confirmed by the ablation experiments. HCformer can combine the advantages of CNN and Transformer, and it has great potential for LDCT image denoising and other tasks.

Regional perception and multi-scale feature fusion network for cardiac segmentation.

Objective.Cardiovascular disease (CVD) is a group of diseases affecting cardiac and blood vessels, and short-axis cardiac magnetic resonance (CMR) images are considered the gold standard for the diagnosis and assessment of CVD. In CMR images, accurate segmentation of cardiac structures (e.g. left ventricle) assists in the parametric quantification of cardiac function. However, the dynamic beating of the heart renders the location of the heart with respect to other tissues difficult to resolve, and the myocardium and its surrounding tissues are similar in grayscale. This makes it challenging to accurately segment the cardiac images. Our goal is to develop a more accurate CMR image segmentation approach.Approach.In this study, we propose a regional perception and multi-scale feature fusion network (RMFNet) for CMR image segmentation. We design two regional perception modules, a window selection transformer (WST) module and a grid extraction transformer (GET) module. The WST module introduces a window selection block to adaptively select the window of interest to perceive information, and a windowed transformer block to enhance global information extraction within each feature window. The WST module enhances the network performance by improving the window of interest. The GET module grids the feature maps to decrease the redundant information in the feature maps and enhances the extraction of latent feature information of the network. The RMFNet further introduces a novel multi-scale feature extraction module to improve the ability to retain detailed information.Main results.The RMFNet is validated with experiments on three cardiac data sets. The results show that the RMFNet outperforms other advanced methods in overall performance. The RMFNet is further validated for generalizability on a multi-organ data set. The results also show that the RMFNet surpasses other comparison methods.Significance.Accurate medical image segmentation can reduce the stress of radiologists and play an important role in image-guided clinical procedures.

3D SAACNet with GBM for the classification of benign and malignant lung nodules.

In view of the low diagnostic accuracy of the current classification methods of benign and malignant pulmonary nodules, this paper proposes a 3D segmentation attention network integrating asymmetric convolution (SAACNet) classification model combined with a gradient boosting machine (GBM). This can make full use of the spatial information of pulmonary nodules. First, the asymmetric convolution (AC) designed in SAACNet can not only strengthen feature extraction but also improve the network's robustness to object flip and rotation detection and improve network performance. Second, the segmentation attention network integrating AC (SAAC) block can effectively extract more fine-grained multiscale spatial information while adaptively recalibrating multidimensional channel attention weights. The SAACNet also uses a dual-path connection for feature reuse, where the model makes full use of features. In addition, this article makes the loss function pay more attention to difficult and misclassified samples by adding adjustment factors. Third, the GBM is used to splice the nodule size, originally cropped nodule pixels, and the depth features learned by SAACNet to improve the prediction accuracy of the overall model. A comprehensive ablation experiment is carried out on the public dataset LUNA16 and compared with other lung nodule classification models. The classification accuracy (ACC) is 95.18%, and the area under the curve (AUC) is 0.977. The results show that this method effectively improves the classification performance of pulmonary nodules. The proposed method has advantages in the classification of benign and malignant pulmonary nodules, and it can effectively assist radiologists in pulmonary nodule classification.

Lightweight YOLOv4 with Multiple Receptive Fields for Detection of Pulmonary Tuberculosis.

The characteristics of pulmonary tuberculosis are complex, and the cost of manual screening is high. The detection model based on convolutional neural network is an essential method for assisted diagnosis with artificial intelligence. However, it also has the disadvantages of complex structure and a large number of parameters, and the detection accuracy needs to be further improved. Therefore, an improved lightweight YOLOv4 pulmonary tuberculosis detection model named MIP-MY is proposed. Firstly, over 300 actual cases are selected to make a common dataset by professional physicians, which is used to evaluate the performance of the model. Subsequently, by introducing the inverted residual channel attention and the pyramid pooling module, a new structure of MIP is created and used as the backbone extractor of MIP-MY, which could further decrease the number of parameters and fuse context information. Then the multiple receptive field module is added after the three effective feature layers of the backbone extractor, which effectively enhances the information extraction ability of the deep feature layer and reduces the miss detection rate of small pulmonary tuberculosis lesions. Finally, the pulmonary tuberculosis detection model MIP-MY with lightweight and multiple receptive field characteristics is constructed by combining each improved modules with multiscale structure. Compared to the original YOLOv4, the model parameters of MIP-MY is reduced by 47%, while the mAP value is raised to 95.32% and the miss detection rate is decreased to 6%. It is verified that the model can effectively assist radiologists in the diagnosis of pulmonary tuberculosis.

Fine-grained calibrated double-attention convolutional network for left ventricular segmentation.

Objective.Left ventricular (LV) segmentation of cardiac magnetic resonance imaging (MRI) is essential for diagnosing and treating the early stage of heart diseases. In convolutional neural networks, the target information of the LV in feature maps may be lost with convolution and max-pooling, particularly at the end of systolic. Fine segmentation of ventricular contour is still a challenge, and it may cause problems with inaccurate calculation of clinical parameters (e.g. ventricular volume). In order to improve the similarity of the neural network output and the target segmentation region, in this paper, a fine-grained calibrated double-attention convolutional network (FCDA-Net) is proposed to finely segment the endocardium and epicardium from ventricular MRI.Approach.FCDA-Nettakes the U-net as the backbone network, and the encoder-decoder structure incorporates a double grouped-attention module that is constructed by a fine calibration spatial attention module (fcSAM) and a fine calibration channel attention module (fcCAM). The double grouped-attention mechanism enhances the expression of information in both spatial and channelwise feature maps to achieve fine calibration.Main Results.The proposed approach is evaluated on the public MICCAI 2009 challenge dataset, and ablation experiments are conducted to demonstrate the effect of each grouped-attention module. Compared with other advanced segmentation methods,FCDA-Netcan obtain better LV segmentation performance.Significance.The LV segmentation results of MRI can be used to perform more accurate quantitative analysis of many essential clinical parameters and it can play an important role in image-guided clinical surgery.

MSANet: Multiscale Aggregation Network Integrating Spatial and Channel Information for Lung Nodule Detection.

Improving the detection accuracy of pulmonary nodules plays an important role in the diagnosis and early treatment of lung cancer. In this paper, a multiscale aggregation network (MSANet), which integrates spatial and channel information, is proposed for 3D pulmonary nodule detection. MSANet is designed to improve the network's ability to extract information and realize multiscale information fusion. First, multiscale aggregation interaction strategies are used to extract multilevel features and avoid feature fusion interference caused by large resolution differences. These strategies can effectively integrate the contextual information of adjacent resolutions and help to detect different sized nodules. Second, the feature extraction module is designed for efficient channel attention and self-calibrated convolutions (ECA-SC) to enhance the interchannel and local spatial information. ECA-SC also recalibrates the features in the feature extraction process, which can realize adaptive learning of feature weights and enhance the information extraction ability of features. Third, the distribution ranking (DR) loss is introduced as the classification loss function to solve the problem of imbalanced data between positive and negative samples. The proposed MSANet is comprehensively compared with other pulmonary nodule detection networks on the LUNA16 dataset, and a CPM score of 0.920 is obtained. The results show that the sensitivity for detecting pulmonary nodules is improved and that the average number of false-positives is effectively reduced. The proposed method has advantages in pulmonary nodule detection and can effectively assist radiologists in pulmonary nodule detection.