Burn injuries are a common traumatic condition, and the early diagnosis of burn depth is crucial for reducing treatment costs and improving survival rates. In recent years, image-based deep learning techniques have been utilized to realize the automation and standardization of burn depth segmentation. However, the scarcity and difficulty in labeling burn data limit the performance of traditional deep learning-based segmentation methods. Mainstream semi-supervised methods face challenges in burn depth segmentation due to single-level perturbations, lack of explicit edge modeling, and ineffective handling of inaccurate predictions in unlabeled data. To address these issues, we propose SBCU-Net, a semi-supervised burn depth segmentation network with contrastive learning and uncertainty correction. Building on the LTB-Net from our previous work, SBCU-Net introduces two additional decoder branches to enhance the consistency between the probability map and soft pseudo-labels under multi-level perturbations. To improve segmentation in complex regions like burn edges, contrastive learning refines the outputs of the three-branch decoder, enabling more discriminative feature representation learning. In addition, an uncertainty correction mechanism weights the consistency loss based on prediction uncertainty, reducing the impact of inaccurate pseudo-labels. Extensive experiments on burn datasets demonstrate that SBCU-Net effectively leverages unlabeled data and achieves superior performance compared to state-of-the-art semi-supervised methods.

As demand grows for real-time diagnostics and control in mirror structure and optical system design, traditional computation-heavy methods face efficiency bottlenecks. To address this, we propose OMS-DNN, a fully data-driven deep learning framework that replaces resource-intensive finite element analysis and ray tracing, while integrating optimization algorithms for forward design. OMS-DNN models complex nonlinear relationships using deep neural networks and captures parameter topologies through a graph convolutional network (GCN). We evaluate OMS-DNN on large-scale datasets, including 20,000 structural and 100,000 Three-Mirror Anastigmat (TMA) optical design samples. It achieves mean absolute percentage errors (MAPE) of 0.98%, 4.09%, and 1.66% across three key metrics. To meet forward design requirements, we integrated the proposed architecture with an optimization algorithm. Experimental results demonstrate that our method can rapidly generate constraint-compliant parameter design solutions tailored to various needs, thereby lowering the entry barrier for designers. These findings validate the effectiveness of OMS-DNN and its potential to transform traditional design workflows.