Accurately segmenting the optic disk (OD) and optic cup (OC) on retinal fundus images is important for treating glaucoma. With the development of deep learning, some CNN-based methods have been implemented to segment OD and OC, but it is difficult to accurately segment OD and OC boundaries affected by blood vessels and the lesion area. To this end, we propose a novel boundary-enhanced adaptive context network (BEAC-Net) for OD and OC segmentation. Firstly, a newly designed efficient boundary pixel attention (EBPA) module enhances pixel-by-pixel feature capture to collect the boundary contextual information of OD and OC in the horizontal and vertical directions. In addition, background noise makes segmenting boundary pixels difficult. To this end, an adaptive context module (ACM) was designed, which simultaneously learns local-range and long-range information to capture richer context. Finally, BEAC-Net adaptively integrates the feature maps from different levels using the attentional feature fusion (AFF) module. In addition, we provide a high-quality retinal fundus image dataset named the 66 Vision-Tech dataset, which advances the field of diagnostic glaucoma. Our proposed BEAC-Net was used to perform extensive experiments on the RIM-ONE-v3, DRISHTI-GS, and 66 Vision-Tech datasets. In particular, BEAC-Net achieved a Dice coefficient of 0.8267 and an IoU of 0.8138 for OD segmentation and a Dice coefficient of 0.8057 and an IoU value of 0.7858 for OC segmentation on the 66 Vision-Tech dataset, achieving state-of-the-art segmentation results.

Efficient and precise colorectal polyp segmentation has significant implications for screening colorectal polyps. Although network variants derived from the Transformer network have high accuracy in segmenting colorectal polyps with complex shapes, they have two main shortcomings: (1) multi-level semantic information at the output of the encoder may result in information loss during the fusion process and (2) failure to adequately suppress background noise during segmentation. To address these challenges, we propose a cross-scale interaction fusion transformer for polyp segmentation (CIFFormer). Firstly, a novel feature supplement module (FSM) supplements the missing details and explores potential features to enhance the feature representations. Additionally, to mitigate the interference of background noise, we designed a cross-scale interactive fusion module (CIFM) that combines feature information between different layers to obtain more multi-scale and discriminative representative features. Furthermore, a boundary-assisted guidance module (BGM) is proposed to help the segmentation network obtain boundary-enhanced details. Extensive experiments on five typical datasets have demonstrated that CIFFormer has an obvious advantage in segmenting polyps. Specifically, CIFFormer achieved an mDice of 0.925 and an mIoU of 0.875 on the Kvasir-SEG dataset, achieving superior segmentation accuracy to competing methods.

Accurate segmentation of tumor PET images has a significance in disease monitoring and clinical treatment. However, the task of accurately segmenting PET images remains challenging due to insufficient access to spatial features of PET images. Therefore, we propose a multiscale attention-guided feature mechanism for 3D tumor segmentation (MA3DSeg) to achieve accurate segmentation of tumor PET images. Specifically, we first utilize a 3D-DX block to construct a lightweight encoder to minimise the redundant information generated across channel features and reduce the model complexity. Additionally, we employ multiscale attention-guided feature enhancement (MAFE) module for 3D networks to fuse channel features and spatial features to fully acquire spatial feature analysis of tumors. Furthermore, we propose explicit and implicit difference information in interaction boundary semantic (IBS) module to enhance the representation of segmentation edges. We conduct rigorous experiments on three publicly available datasets ECPC-IDS, Hecktor 2022 and AutoPET, which demonstrated superior performance than other competitive networks. MA3DSeg significantly improves Dice by 0.5\(\%\) on Hecktor 2022 and RVD by 2.1\(\%\) on ECPC-IDS compared with 3D UX-Net. The experimental results show that MA3DSeg achieves excellent tumor segmentation performance and generalization ability.

The acoustic spectrogram arranges the frequencies in the sound along the frequency spread, and translates the spectral changes into the intensity, wavelength and frequency of the electrical signals. Currently, the extensive use of convolutional neural networks for spectral image classification can extract signal features in the spectrogram, but the redundancy of noisy data generated by a large number of bands of the spectrum affects the feature information at different levels of the image. In order to optimize this problem, this paper proposes a multi-branch residual-connected Efficient Global Attention (EGA) acoustic spectral image classification network based on the attention mechanism, which firstly separates the components with their respective acoustic features from the spectral noise, so as to achieve the purpose of noise reduction, and then extracts the Phase Resolved Partial Discharge (PRPD) Spectrum of the Intermediate Frequency (IF) cycle for the original signals that have undergone noise reduction, which is based on the attention mechanism through the Improved Global Attention Mechanism (IGAM) in the EGA of the backbone network. mechanism pays more attention to the channel and spatial features of the spectrogram, then improves the feature extraction ability by residual connection, and finally performs feature fusion with the mask branch. The results show that a more accurate detection of abnormal partial discharge type of carbon brushes in gantry cranes is made, and the feasibility and innovativeness of the method is verified through experiments and production use.