Ying Zhao

Recently, semi-supervised learning has demonstrated significant potential in the field of medical image segmentation. However, the majority of the methods fail to establish connections among diverse sample data. Moreover, segmentation networks that utilize fixed parameters can impede model training and even amplify the risk of overfitting. To address these challenges, this paper proposes an adversarial consistency-based semi-supervised segmentation method, leveraging a dual multiscale mean teacher model. First, by designing a discriminator network with adaptive feature selection and training it alternately with the segmentation network, the method enhances the segmentation network's ability to transfer knowledge from the limited labeled data to the unlabeled data. The discriminator evaluates the quality of the segmentation network's results for both labeled and unlabeled data, while simultaneously guiding the network to learn consistency in segmentation performance throughout the training process. Second, we design a Triple-attention dynamic convolutional (TADC) module, which allows the convolution kernel parameters to be adjusted flexibly according to different input data. This improves the feature representation capability of the network model and helps reduce the risk of overfitting. Finally, we propose a novel feature selection and fusion module (FSFM) within the segmentation network, which dynamically selects and integrates important features to enhance the saliency of key information, improving the overall performance of the model. The proposed adversarial consistency-based semi-supervised segmentation method is applied to the MosMedData dataset. The results demonstrate that the segmentation network outperforms the baseline model, achieving improvements of 3.83%, 3.97%, 3.14% in terms of Dice, Jaccard, and NSD scores, respectively, for the segmentation of pneumonia lesions. The proposed segmentation method outperforms state-of-the-art segmentation networks and demonstrates superior potential for segmenting pneumonia lesions, as evidenced by extensive experiments conducted on the MosMedData and COVID-19-P20 datasets.

Introduction: This study aims to propose and evaluate a two-stage semi-supervised segmentation framework with dual multiscale uncertainty estimation and graph reasoning, addressing the challenges of obtaining high-precision pixel-level labels and effectively utilizing unlabeled data for accurate pneumonia lesion segmentation. Methods: First, we design a guided supervised training strategy for modeling aleatoric uncertainty (AU) at dual scales, reducing the impact on segmentation performance caused by aleatoric uncertainties introduced by blurred lesions and their boundaries in the image. Second, we design a training strategy for multi-scale noisy pseudo-label correction to reduce the cognitive bias problem caused by unreliable predictions in the model. Finally, we design a new combination of fused feature interaction graph reasoning (FIGR) and attention modules, which enables the network model to better capture image features in small infected regions. Results: Our study was validated using the MosMedData public dataset. The proposed algorithm improves the performance by 1.25%, 1.03%, 2.98%, and 0.59% on Dice, Jaccard, normalized surface dice (NSD), and average distance of boundaries (ADB), respectively, compared to the baseline model. Discussion: Our semi-supervised pneumonia segmentation framework, through two-stage multi-scale uncertainty estimation and modeling, significantly improves segmentation performance by leveraging unlabeled data and addressing uncertainties, offering clinical benefits in pneumonia diagnosis while facing challenges in generalization and computational efficiency that future work will target with GAN-based data synthesis and architecture optimization. Conclusion: It can be convincingly concluded that the proposed algorithm is of profound importance and value in the domain of clinical practice.

In recent years, filter pruning has been recognized as an indispensable technique for mitigating the significant computational complexity and parameter burden associated with deep convolutional neural networks (CNNs). To date, existing methods are based on heuristically designed pruning metrics or implementing weight regulations to penalize filter parameters during the training process. Nevertheless, human-crafted pruning criteria tend not to identify the most critical filters, and the introduction of weight constraints can inadvertently interfere with weight training. To rectify these obstacles, this article introduces a novel mask learning method for autonomous filter pruning, negating requirements for weight penalties. Specifically, we attribute a learnable mask to each filter. During forward propagation, the mask is transformed to a binary value of 1 or 0, serving as indicators for the necessity of corresponding filter pruning. In contrast, throughout backward propagation, we use straight-through estimator (STE) to estimate the gradient of masks, accommodating the non-differentiable characteristic of the rounding function. We verify that these learned masks aptly reflect the significance of corresponding filters. Concurrently, throughout the mask learning process, the training of neural network parameters remains uninfluenced, therefore protecting the normal training process of weights. The efficacy of our proposed filter pruning method based on mask learning, termed MLPruner, is substantiated through its application to prevalent CNNs across numerous representative benchmarks.

Chaotic system plays a momentous role in the field of image encryption. An encryption algorithm based on Long Short-term Memory optimized by Kolmogorov–Arnold Networks (LSTM-KAN) for generating a two-dimensions new chaotic signal is proposed in this paper. The chaotic behavior of the signal is demonstrated through a series of analytical methods. The advantage of the new chaotic signal is that it has no formula expression. Following this, a new image encryption algorithm is subsequently proposed. The algorithm divides the image into high and low bit planes, scrambles them using Hilbert–Zigzag. After performing a confusion operation with the chaotic signal, it selectively shifts the 8-bit binary pixels and subsequently performs global diffusion. The simulation and analysis results show that our image encryption algorithm can resist various attacks and offers higher security. Especially, it performs excellently against differential attack.

In order to improve the perception of moving objects for visual prosthesis implanters, a strategy of motion target optimization based on video instance segmentation under simulated prosthesis vision was proposed. Psychophysical experiments based on the MaskTrack-rcnn algorithm were performed on three optimization strategies( DP, Mr and Ours) at five resolutions(16×16, 24×24, 32×32, 48×48 and 64×64). The experimental results showed, the method proposed in this paper has a significant reduction in recognition time and a certain degree of significant improvement in recognition accuracy compared with DP and Mr under these five resolutions. It was verified to improve subjects' perceptual recognition of moving target objects in daily life.

Objective: False-positive nodule reduction is a crucial part of a computer-aided detection (CADe) system, which assists radiologists in accurate lung nodule detection. In this research, a novel scheme using multi-level 3D DenseNet framework is proposed to implement false-positive nodule reduction task. Methods: Multi-level 3D DenseNet models were extended to differentiate lung nodules from falsepositive nodules. First, different models were fed with 3D cubes with different sizes for encoding multi-level contextual information to meet the challenges of the large variations of lung nodules. In addition, image rotation and flipping were utilized to upsample positive samples which consisted of a positive sample set. Furthermore, the 3D DenseNets were designed to keep low-level information of nodules, as densely connected structures in DenseNet can reuse features of lung nodules and then boost feature propagation. Finally, the optimal weighted linear combination of all model scores obtained the best classification result in this research. Results: The proposed method was evaluated with LUNA16 dataset which contained 888 thin-slice CT scans. The performance was validated via 10-fold cross-validation. Both the Free-response Receiver Operating Characteristic (FROC) curve and the Competition Performance Metric (CPM) score show that the proposed scheme can achieve a satisfactory detection performance in the falsepositive reduction track of the LUNA16 challenge. Conclusion: The result shows that the proposed scheme can be significant for false-positive nodule reduction task.