Improved U-Net3+ with stage residual for brain tumor segmentation.

BACKGROUND: For the encoding part of U-Net3+,the ability of brain tumor feature extraction is insufficient, as a result, the features can not be fused well during up-sampling, and the accuracy of segmentation will reduce. METHODS: In this study, we put forward an improved U-Net3+ segmentation network based on stage residual. In the encoder part, the encoder based on the stage residual structure is used to solve the vanishing gradient problem caused by the increasing in network depth, and enhances the feature extraction ability of the encoder which is instrumental in full feature fusion when up-sampling in the network. What's more, we replaced batch normalization (BN) layer with filter response normalization (FRN) layer to eliminate batch size impact on the network. Based on the improved U-Net3+ two-dimensional (2D) model with stage residual, IResUnet3+ three-dimensional (3D) model is constructed. We propose appropriate methods to deal with 3D data, which achieve accurate segmentation of the 3D network. RESULTS: The experimental results showed that: the sensitivity of WT, TC, and ET increased by 1.34%, 4.6%, and 8.44%, respectively. And the Dice coefficients of ET and WT were further increased by 3.43% and 1.03%, respectively. To facilitate further research, source code can be found at: https://github.com/YuOnlyLookOne/IResUnet3Plus . CONCLUSION: The improved network has a significant improvement in the segmentation task of the brain tumor BraTS2018 dataset, compared with the classical networks u-net, v-net, resunet and u-net3+, the proposed network has smaller parameters and significantly improved accuracy.

[Research on three-dimensional skull repair by combining residual and informer attention].

Cranial defects may result from clinical brain tumor surgery or accidental trauma. The defect skulls require hand-designed skull implants to repair. The edge of the skull implant needs to be accurately matched to the boundary of the skull wound with various defects. For the manual design of cranial implants, it is time-consuming and technically demanding, and the accuracy is low. Therefore, an informer residual attention U-Net (IRA-Unet) for the automatic design of three-dimensional (3D) skull implants was proposed in this paper. Informer was applied from the field of natural language processing to the field of computer vision for attention extraction. Informer attention can extract attention and make the model focus more on the location of the skull defect. Informer attention can also reduce the computation and parameter count from N 2 to log( N). Furthermore,the informer residual attention is constructed. The informer attention and the residual are combined and placed in the position of the model close to the output layer. Thus, the model can select and synthesize the global receptive field and local information to improve the model accuracy and speed up the model convergence. In this paper, the open data set of the AutoImplant 2020 was used for training and testing, and the effects of direct and indirect acquisition of skull implants on the results were compared and analyzed in the experimental part. The experimental results show that the performance of the model is robust on the test set of 110 cases fromAutoImplant 2020. The Dice coefficient and Hausdorff distance are 0.940 4 and 3.686 6, respectively. The proposed model reduces the resources required to run the model while maintaining the accuracy of the cranial implant shape, and effectively assists the surgeon in automating the design of efficient cranial repair, thereby improving the quality of the patient's postoperative recovery.

Dual adversarial models with cross-coordination consistency constraint for domain adaption in brain tumor segmentation.

The brain tumor segmentation task with different domains remains a major challenge because tumors of different grades and severities may show different distributions, limiting the ability of a single segmentation model to label such tumors. Semi-supervised models (e.g., mean teacher) are strong unsupervised domain-adaptation learners. However, one of the main drawbacks of using a mean teacher is that given a large number of iterations, the teacher model weights converge to those of the student model, and any biased and unstable predictions are carried over to the student. In this article, we proposed a novel unsupervised domain-adaptation framework for the brain tumor segmentation task, which uses dual student and adversarial training techniques to effectively tackle domain shift with MR images. In this study, the adversarial strategy and consistency constraint for each student can align the feature representation on the source and target domains. Furthermore, we introduced the cross-coordination constraint for the target domain data to constrain the models to produce more confident predictions. We validated our framework on the cross-subtype and cross-modality tasks in brain tumor segmentation and achieved better performance than the current unsupervised domain-adaptation and semi-supervised frameworks.

Dynamically aggregating MLPs and CNNs for skin lesion segmentation with geometry regularization.

BACKGROUND AND OBJECTIVE: Melanoma is a highly malignant skin tumor. Accurate segmentation of skin lesions from dermoscopy images is pivotal for computer-aided diagnosis of melanoma. However, blurred lesion boundaries, variable lesion shapes, and other interference factors pose a challenge in this regard. METHODS: This work proposes a novel framework called CFF-Net (Cross Feature Fusion Network) for supervised skin lesion segmentation. The encoder of the network includes dual branches, where the CNNs branch aims to extract rich local features while MLPs branch is used to establish both the global-spatial-dependencies and global-channel-dependencies for precise delineation of skin lesions. Besides, a feature-interaction module between two branches is designed for strengthening the feature representation by allowing dynamic exchange of spatial and channel information, so as to retain more spatial details and inhibit irrelevant noise. Moreover, an auxiliary prediction task is introduced to learn the global geometric information, highlighting the boundary of the skin lesion. RESULTS: Comprehensive experiments using four publicly available skin lesion datasets (i.e., ISIC 2018, ISIC 2017, ISIC 2016, and PH2) indicated that CFF-Net outperformed the state-of-the-art models. In particular, CFF-Net greatly increased the average Jaccard Index score from 79.71% to 81.86% in ISIC 2018, from 78.03% to 80.21% in ISIC 2017, from 82.58% to 85.38% in ISIC 2016, and from 84.18% to 89.71% in PH2 compared with U-Net. Ablation studies demonstrated the effectiveness of each proposed component. Cross-validation experiments in ISIC 2018 and PH2 datasets verified the generalizability of CFF-Net under different skin lesion data distributions. Finally, comparison experiments using three public datasets demonstrated the superior performance of our model. CONCLUSION: The proposed CFF-Net performed well in four public skin lesion datasets, especially for challenging cases with blurred edges of skin lesions and low contrast between skin lesions and background. CFF-Net can be employed for other segmentation tasks with better prediction and more accurate delineation of boundaries.

MAD-Net: Multi-attention dense network for functional bone marrow segmentation.

Radiotherapy is the main treatment modality for various pelvic malignancies. However, high intensity radiation can damage the functional bone marrow (FBM), resulting in hematological toxicity (HT). Accurate identification and protection of the FBM during radiotherapy planning can reduce pelvic HT. The traditional manual method for contouring the FBM is time-consuming and laborious. Therefore, development of an efficient and accurate automatic segmentation mode can provide a distinct leverage in clinical settings. In this paper, we propose the first network for performing the FBM segmentation task, which is referred to as the multi-attention dense network (named MAD-Net). Primarily, we introduce the dense convolution block to promote the gradient flow in the network as well as incite feature reuse. Next, a novel slide-window attention module is proposed to emphasize long-range dependencies and exploit interdependencies between features. Finally, we design a residual-dual attention module as the bottleneck layer, which further aggregates useful spatial details and explores intra-class responsiveness of high-level features. In this work, we conduct extensive experiments on our dataset of 3838 two-dimensional pelvic slices. Experimental results demonstrate that the proposed MAD-Net transcends previous state-of-the-art models in various metrics. In addition, the contributions of the proposed components are verified by ablation analysis, and we conduct experiments on three other datasets to manifest the generalizability of MAD-Net.

Joint Dense Residual and Recurrent Attention Network for DCE-MRI Breast Tumor Segmentation.

Breast cancer detection largely relies on imaging characteristics and the ability of clinicians to easily and quickly identify potential lesions. Magnetic resonance imaging (MRI) of breast tumors has recently shown great promise for enabling the automatic identification of breast tumors. Nevertheless, state-of-the-art MRI-based algorithms utilizing deep learning techniques are still limited in their ability to accurately separate tumor and healthy tissue. Therefore, in the current work, we propose an automatic and accurate two-stage U-Net-based segmentation framework for breast tumor detection using dynamic contrast-enhanced MRI (DCE-MRI). This framework was evaluated using T2-weighted MRI data from 160 breast tumor cases, and its performance was compared with that of the standard U-Net model. In the first stage of the proposed framework, a refined U-Net model was utilized to automatically delineate a breast region of interest (ROI) from the surrounding healthy tissue. Importantly, this automatic segmentation step reduced the impact of the background chest tissue on breast tumors' identification. For the second stage, we employed an improved U-Net model that combined a dense residual module based on dilated convolution with a recurrent attention module. This model was used to accurately and automatically segment the tumor tissue from healthy tissue in the breast ROI derived in the previous step. Overall, compared to the U-Net model, the proposed technique exhibited increases in the Dice similarity coefficient, Jaccard similarity, positive predictive value, sensitivity, and Hausdorff distance of 3%, 3%, 3%, 2%, and 16.2, respectively. The proposed model may in the future aid in the clinical diagnosis of breast cancer lesions and help guide individualized patient treatment.

Research on multi-path dense networks for MRI spinal segmentation.

Accurate and robust segmentation of anatomical structures from magnetic resonance images is valuable in many computer-aided clinical tasks. Traditional codec networks are not satisfactory because of their low accuracy of edge segmentation, the low recognition rate of the target, and loss of detailed information. To address these problems, this study proposes a series of improved models for semantic segmentation and progressively optimizes them from the three aspects of convolution module, codec unit, and feature fusion. Instead of the standard convolution structure, we apply a new type of convolution module for the feature extraction. The networks integrate a multi-path method to obtain richer-detail edge information. Finally, a dense network is utilized to strengthen the ability of the feature fusion and integrate more different-level information. The evaluation of the Accuracy, Dice coefficient, and Jaccard index led to values of 0.9855, 0.9185, and 0.8507, respectively. These metrics of the best network increased by 1.0%, 4.0%, and 6.1%, respectively. Boundary F1-Score reached 0.9124 indicating that the proposed networks can segment smaller targets to obtain smoother edges. Our methods obtain more key information than traditional methods and achieve superiority in segmentation performance.